《OpenAI：2025歐盟經濟藍圖（英文版）（21頁）.pdf》由會員分享，可在線閱讀，更多相關《OpenAI：2025歐盟經濟藍圖（英文版）（21頁）.pdf（21頁珍藏版）》請在三個皮匠報告上搜索。

1、EU Economic BlueprintOpenAI in EuropeEU Economic Blueprint2Second PrincipleEnsure that EU rules are streamlined and work in sync to enable AI progress rather than hinder it.Third PrincipleMaximize the AI opportunity through widespread adoption across all sectors,regions and society.Fourth PrincipleE

2、nsure that AI is built responsibly and reflects European values.ConclusionIntroductionThe Economics of AI ProgressDeveloping Responsible AIFirst PrincipleEstablish and grow the foundations needed for sustained AI growth-chips,data,energy and talent.0306070811131720OpenAI in EuropeEU Economic Bluepri

3、nt3IntroductionOpenAIs mission is to ensure that artificial intelligence benefits everyone.To us,that means building AI that helps people solve hard problems,because by helping with the hard problems,AI can benefit the most people possible through better healthcare and education,more scientific disc

4、overies,better public services,new forms of creative expression,and increased economic growth and productivity.Were off to a strong start,creating freely available intelligence being used by half a billion people around the world on a weekly basis to ideate,discover,and innovate beyond what were cur

5、rently capable of doing on our own.Across the European Union,our tools are driving progress.Existing partnerships are helping to:Speed up development of life-saving medical treatmentsAccelerate scientific researchImprove AI access for underserved communitiesImprove educationUnlock new forms of creat

6、ivity for artistsPowering a new generation of start ups and entrepreneursHelping researchers design new drugs faster than before,including through our partnership with Sanofi.At leading European laboratories and universities like Science Po in France and the Max Planck Society in Germany.By increasi

7、ng accessibility through Danish start-up Be My Eyes,and improving economic and social inclusion with Simplon in France for vulnerable populations.Including bringing AI to schools across Estonia and enabling teachers at ESCP Business School to create personalized lesson plans,spend more time with stu

8、dents,and safely adopt AI in the classroom to improve learning outcomes.By increasing accessibility of artistic production,enabling creators to develop content in innovative ways,like immersive experiences at Versailles and content translations with Spotify.Such as Synthflow,Pigment and Brainly that

9、 are using OpenAIs technology to compete globally with AI-driven services.Outside of the US,Germany is the country with the largest number of API developers building on OpenAIs technology.OpenAI in EuropeEU Economic Blueprint4We believe the EU needs to act more boldly and decisively to maximize AIs

10、possibilities while also ensuring its used responsibly to mitigate potential negative effects.The oppor-tunity AI presents to spur productivity and revive the EUs economic competitiveness is too compelling to forfeit.At the same time,AI is too powerful a technology to be led and shaped by autocratic

11、 governments who dont share our common values yet that is a growing risk if democracies dont move with speed and determination,both to build AI on democratic principles and to demonstrate its benefits.Prosperity,broadly shared,is a cornerstone and founding principle of European integration and progr

12、ess.Shared prosperity from AI is as near and measurable as the new jobs and growth that will come from building the needed infrastructure for this digital age.Soon,as our CEO Sam Altman has written,AI will help our children do things we cant.Not far off is a future in which everyones life can be bet

13、ter than anyones life is today.With such opportunities in sight,we want to work with the EU policy community to ensure that AIs benefits are shared responsibly and equitably across the Union.This Blueprint champions the entrepreneurial spirit,scientific excellence,and individual freedoms at the hear

14、t of Europe,and reflects the EUs commitment to social inclusion and fundamental rights.It has been shaped by the conviction that Europe can best guard and prosper from its values through accelerating technological progress.Just as forward-looking EU leaders once chose cooperation and innovation to b

15、uild the single market,now they must foster an environment where AI can thrive responsibly and create real incentives for Europeans to adapt to and adopt AI.The EU and its member states working together with academia,civil society and industry can pave the road for AI built by Europe,in Europe,for E

16、urope.We want to be a thought partner to the EU on this journey and are sharing four principles that we believe will be vital to the EUs success:01030204Establish and grow the foundations needed for sustained AI growth-chips,data,energy and talent.Maximize the AI opportunity through widespread adopt

17、ion across all sectors,regions and societyEnsure that EU rules are streamlined and work in sync to enable AI progress rather than hinder it.Ensure that AI is built responsibly and reflects European values.OpenAI in EuropeEU Economic Blueprint5The EUs strength is its scale-concerted effort across the

18、se four principles with balanced,harmonized rules can spur competition and innovation,help native AI companies start and scale,and ensure that the benefits of AI reach all Europeans across the 27 member states.A thread between these principles is the need for the EU to simplify its overall approach

19、and ensure its regulations and initiatives are working in sync-creating a flywheel for AI growth rather than nurturing adoption and innovation in some areas while hindering it in others.We call for EU policymakers and regulators to reset and rethink their current approach to regulating AI in order t

20、o review the accumulated rules that former European Central Bank President Mario Draghi has called a drag on European competitiveness.We believe policy-makers should evaluate which rules are boosting the EUs AI sector and should be kept in place,and which rules are holding it back and should be revi

21、sed.We commend the intent to bring forward a Digital Simplification Package as part of the broader simplification agenda and urge policymakers to consider how it can best serve the AI sector.We also believe that the forthcoming AI Continent and Apply AI strategies will be decisive and should gather

22、strong support across the political spectrum and maintain a rigorous focus on increasing AI adoption and AI-driven opportunities.This Blueprint a living document we will update as we learn more from our partnership is intended to sync with whats working for the EUs AI sector and encourage AI rulemak

23、ing that ignites productivity and growth.We are engaging with regulators and leaders across the EU to share our technological advances and how they can drive economic growth and meet local expectations.Its also why we continue to expand our presence through our offices in Dublin,Paris,Brussels and s

24、oon,in Munich,and why were partnering with local institutions to ensure that we bring AIs economic benefits to Europeans from all walks of life.We look forward to building on this foundation for collaboration and advancing democratic AI with you.Chris Lehane Chief Global Affairs OfficerSandro Gianel

25、la Head of EMEA Public Policy&PartnershipsOpenAI in EuropeEU Economic Blueprint6The Economics of AI ProgressInvesting more in AI will continue to make it more capable and beneficial.The intelligence of an AI model roughly equals the log of the resources used to train and run it.Until recently,scalin

26、g progress has primarily come from training compute and data,but we have shown how to make intelligence scale from inference compute,as well.The scaling laws that pre-dict these gains are incredibly precise over many orders of magnitude.It follows that further investment will lead to further gains,a

27、nd further benefits to society:We believe that the socioeconomic value of linearly increasing intelligence is super-exponential in nature.The cost to use a given level of AI capability falls by about 10 x every 12 months,and lower prices lead to much more use.We saw this in the change in token cost

28、between GPT-4 in early 2023 and GPT-4o in mid-2024,where the price per token dropped about 150 x in that time period.Moores Law predicted that the number of transistors on a microchip would double roughly every two years;the decrease in the cost of using AI is even more dramatic.The amount of time i

29、t takes to improve an AI model keeps decreasing.Put another way,AI models are catching up with human intelligence at an increasing rate.The typical time it takes for a computer to beat humans at a given benchmark has fallen from 20 years after the benchmark was introduced,to five years,and now to on

30、e to two yearsand we see no reason why those advances will stop in the near future.Our work at OpenAI suggests that as AI advances,progress accelerates and becomes increasingly affordable:010203OpenAI in EuropeEU Economic Blueprint7Developing Responsible AIOpenAIs Preparedness Framework is the safet

31、y protocol guiding how we manage the largest-scale risks when developing and deploying our increasingly advanced AI models.It helps us measure and protect against the risk of severe harm from dual-use frontier capabilities.We are constantly refining our practices and advancing the science underpinni

32、ng these developments,to unlock the benefits of these technologies while addressing their risks.OpenAIs Model Spec details our approach to shaping the behavior of AI models to ensure they are useful,safe,and aligned with both user and developer needs.The Spec provides concrete guidelines for mitigat

33、ing risks like misaligned goals,execution errors,and harmful instructions,ensuring our models ask clarifying questions when needed and follow a structured hierarchy in decision-making.By publishing this document under the Creative Commons CC0 1.0 deed,we invite public discussion and collaboration,re

34、inforcing our commitment to transparency and continuous improvement in AI safety and alignment.At OpenAI,we believe in Privacy for the Intelligence Agea commitment to safeguarding personal data through implementing robust data minimization measures,meaningful transparency,and giving users control ov

35、er their data.We pioneered and embedded state-of-the-art privacy safeguards throughout the entire AI lifecyclefrom pre-training to deploymentensuring that privacy is a foundational element of how we build and deploy our technologies.At OpenAI,we strive to ensure that our AI technologies align with h

36、uman values and operate securely.Our commitment to transparency,accountability,and thorough testing helps mitigate risks and maximize societal benefits,setting a high standard for safe AI development.010203OpenAI in EuropeEU Economic Blueprint8For the EU,the AI era is an unmissable opportunity to dr

37、ive growth and bolster its strategic capabilities.In the AI era,infrastructure is destiny.Successful nations will turn resources into competitive advantages and in this AI age,the critical resources are compute,data,energy,and talent.Marshalling these in the EU through bold investment will create wi

38、de-spread economic opportunity,and thoughtful harmonized rules will serve as a complement,mitigating risks to citizens and enabling the region to capitalize on its innovative potential.Establish and grow the foundations needed for sustained AI growth-chips,data,energy and talent.We believe the EU sh

39、ould seize the moment and make bold investments so that it has the physical and human infrastructure to support AI at scale.In concrete terms,Europe must be able to produce and access world-class computing power,abundant clean energy,vast and diverse datasets,and a skilled workforce.In the past,Euro

40、pes economic strength was built on infrastructure like highways,factories,and universities;in the AI future,it will be about supercomputers,fiber optic networks,power grids,and research centers.If Europe lags in any of these areas,it could find itself dependent on external providers for critical cap

41、abilities something the EU has recognized as a strategic risk.ComputeAbundant and affordable compute will be the engine of AI development in the EU.The good news is that steps are already being taken including recent plans to mobilize 200 billion for infrastructure across the EU and 109 billion in i

42、nvestment in France.These come on top of existing EuroHPC and AI Factory initiatives,which are laudable steps to build some of the worlds fastest supercomputers and make them available to researchers and startups.Europes challenge is now to accelerate and expand these initiatives.To complement AI fa

43、ctories initiatives,we recommend an AI Compute Scaling Plan that boosts EU comput-ing capacity by at least 300%by 2030,prioritizing low-latency,geographically spread inference infrastructure.Access to chips is vital to drive down the cost of compute and help make it abundant in Europe.Our recent sub

44、mission to the White House Office of Science and Technology for the upcoming US AI Action Plan,recommended that countries that commit to democratic AI principles by deploying AI systems in ways that promote more freedoms for their citizens could be considered Tier I countries.Tier II status should b

45、e limited to only those countries that have a history of failing to prevent export-controlled chips and other US-developed IP from being diverted into,or used by Tier III countries.Critical ResourcesFirst PrincipleOpenAI in EuropeEU Economic Blueprint9DataIn AI,data is the raw material that fuels AI

46、 model training.The EU has vast amounts of data(from scientific research to industrial data to public sector information),but much of it remains siloed,or underutilized.To innovate in AI,developers need access to rich,diverse,and relevant datasets while respecting privacy,data protection and securit

47、y.The EU should assure AI developers and researchers that their models will have the ability to learn from publicly available sources and that more data will be made available in respon-sible ways.This would also attract infrastructure investment because the more the EU is able to reduce barriers to

48、 data access,the more it will be able to convince investors and companies to place their multi-billion euro data infrastructures in its jurisdiction.EnergyFor the EU to support widespread AI,it must ensure sufficient,sustainable energy supply.This aligns well with Europes green objectives:investment

49、s in clean energy(solar,wind,nuclear,and emerging sources like fusion)serve the dual purpose of fighting climate change and powering digital growth.We recommend dramatically increasing European capacity for renewable energy generation and modernizing power grids.Streamlining permitting for new energ

50、y projects and cross-border transmission lines(part of the EUs existing efforts under the Trans-European Networks for Energy)will help meet AIs electricity needs in a climate-friendly way.TalentEven with the best hardware and data,AI leadership ultimately depends on people researchers,engineers,doma

51、in experts,and informed users-and Europe must focus on retaining and cultivating its considerable human capital.To do this,the EU should significantly expand investments in STEM education,specialized AI scholarships akin to an“AI Erasmus,”and networks of AI research excellence centers across the con

52、tinent.Equally crucial is large-scale workforce reskilling through apprenticeships and vocational programs,enabling current workers to adapt their skills for AI-driven roles.Broad and equitable access to AI education from primary school onward,similar to Estonias suc-cessful integration of AI in cur

53、ricula,can prepare Europeans for an increasingly AI-centric world and reinforce Europes digital values.Additionally,Europe should attract global AI talent by improving policies like the EU Blue Card and investing in interdisciplinary expertise to ensure responsible and ethical AI deployment.OpenAI i

54、n EuropeEU Economic Blueprint10Our policy proposals AI Compute Scaling Plan:Increase the EUs computing capacity by at least 300%by 2030,with a clear focus on low-latency,geographically-distributed infrastructure optimized for inference,as a complement to the AI factories focused on training.Green AI

55、 Grid:Fast-track renewable energy projects,prioritizing streamlined approvals for data centers,to deliver carbon-neutral AI infrastructure EU-wide by 2030.EU AI Data Spaces:Implement sector-specific AI data spaces(health,industrial,environmental,public)under the EU Data Strategy by 2027 to foster tr

56、usted data sharing;advancing AI through data sets with privacy-preserving safeguards in place,foster transparency,and accelerate safe innovation.Public Data Availability:Make the availability of online public data a top priority to fuel AI innovation and model development that is reflective of Europ

57、es diverse languages and cultures.OpenAI in EuropeEU Economic Blueprint11AI is a rapidly evolving technology poised to become integral across society,requiring thoughtful regulation to mitigate risks without stifling innovation.Unfortunately,today Europes regulatory environment is often too complex

58、as Mario Draghi noted in his report,the EU now has around 100 tech-related laws and over 270 bodies regulating digital sectors across member states.EU policymakers should focus on reducing barriers within the single market by simplifying the digital rulebook and eliminating redundant or obsolete pro

59、posals.The ongoing Omnibus Simplification Package and broader simplifi-cation agenda promoted by the European Commission represent critical opportunities to streamline regulations and strengthen EU competitiveness.Ensure that EU rules are stream-lined and work in sync to enable AI progress rather th

60、an hinder it.AI ActAny discussion of AI regulation in Europe has to address the AI Act,the worlds first com-prehensive framework for AI.OpenAI supports the AI Acts core objective to ensure AI systems are safe and we have engaged constructively with EU lawmakers throughout its drafting and implementa

61、tion.It is crucial that this Act strikes the right balance:pro-tecting citizens and giving certainty to businesses,while not unduly burdening innovative developers or pushing research out of Europe.At OpenAI,weve already taken steps to adopt robust safety measures in our operations.Weve been one of

62、the first AI companies to design a detailed framework that guides the development and deployment of our models,including rigorous testing and the publication of safety documentation for each model.We actively engage with global and European initiatives on AI governance from working with law enforcem

63、ent on misuse cases to participating in multistakeholder forums on AI ethics.Were also developing tools to help users and third parties identify AI-generated content(through watermarking techniques and other research),contributing to the broader transparency goal.With an approach that strikes a bala

64、nce between common-sense guards against risks,and enabling and incentivising innovation,Europe can both protect its citizens and societies while also seizing what AI offers.SimplificationA house divided against itself cannot stand,and nor can a strategy that nurtures AI growth in some areas while hi

65、ndering adoption in others.Second PrincipleOpenAI in EuropeEU Economic Blueprint12The Draghi report was right that the sheer breadth and quantity of EU regulations hamper innovation,slow economic growth,and pose an“existential challenge”to the European Unions future.They close the door on the EUs ab

66、ility to build AI that is of,by,and for Europe.Now,as policymakers take a step back to evaluate which rules strengthen the EUs AI sector and should be preserved,versus which ones are holding it back and should notthey should make sure theyre considering the full picture.This is a one-way door decisi

67、on that will have huge consequences for Europes future progress and prosperity.If policymakers get it right,it will open the door to building AI for Europeand unlocking the progress and prosperity gains the technology offers.If they dont,Europe risks falling further behind in the global competition

68、over AIs future.HarmonizationThe EU must be vigilant against the risks posed by regulatory fragmentation.EU rules should apply clearly and consistently across the single market.Europes abundant talent and innovative spirit remain constrained by national barriers,limiting the seamless,cross-bor-der g

69、rowth that is essential for global competitiveness.The EUs strength is its scale:one set of balanced rules for 27 countries will spur competition and innovation,help native AI companies start and scale,and ensure that the benefits of AI reach all Europeans.For example,a pan-European entity for start

70、ups is an area that would benefit from a unified regulatory vision.As underscored by Ursula von der Leyen,while startups in the US benefit from a cohesive market,European companies struggle with fragmented ecosystems.Establishing a unified legal frameworkas put forward by EU Incwill empower startups

71、 to secure capital more effectively and realize their potential across and beyond Europe.Our policy proposals AI Accelerator Fund:Launch a dedicated 1 billion fund to rapidly finance pilot projects demonstrating clear societal or economic value of AI.A European AI Readiness Index:To encourage nation

72、al action,publish a yearly ranking of EU countries based on a methodology assessing AI adoption,skills,infrastructure,and regulation readiness.National AI Readiness Officers:By 2027,each EU member state should appoint a dedicated“AI Readiness Officer,”responsible for accelerating national AI adoptio

73、n and creating synergies among member states.Pan-European Start-up Entity:By 2026,launch a unified legal framework that harmonizes corporate law,taxation,and compliance across the EU,enabling startups to operate seamlessly.OpenAI in EuropeEU Economic Blueprint13As AI models and systems become more c

74、apable and versatile,they will increasingly bring prosperity and advantages to people and organizations.Yet translating the potential of AI into concrete economic and social benefits will not happen automatically.These benefits will not materialize without adequate AI adoption at the organizational

75、level and AI literacy at the individual level.To us,this means that beyond the foundational investments needed in research and in infrastructure,we believe that governments and institutions in the EU,both nationally and locally,have a central role to play in incentivizing adoption in the public and

76、private sector.Moreover,preparing,educating and upskilling individuals would have a high impact on the development of AI throughout the continent.Europe has enormous assets to build on,including world-class universities and research labs,a highly skilled workforce,and leading industries in sectors l

77、ike manufacturing,automotive,pharmaceuticals,and finance.By infusing AI into these sectors,Europe can supercharge productivity and create new high-value jobs.Maximize the AI opportunity through widespread adoption across all sectors,regions and society.Private SectorEuropean companies must leverage

78、AI to advance their industries.By encouraging such partnerships and innovation,the EU can ensure that its single market of over 450 million people becomes one of the worlds most attractive arenas for AI-driven business.Every European sector from agriculture and tourism,to energy and creative industr

79、ies stands to gain from AI adoption,whether through cost savings,better services,or entirely new and innovative AI-based offerings.In healthcare,OpenAI is partnering with Sanofi to use AI for accelerating clinical trials and drug discovery,helping life-saving treatments reach patients faster.In tele

80、communications,Orange is fine-tuning our open-source speech model to better understand African regional languages,illustrating how European companies can localize and build on AI to serve global markets.Meanwhile,T-Mobile is working with OpenAI to launch IntentCX,a platform that uses generative AI t

81、o transform how businesses handle customer interactionsimproving service quality,reducing response times,and empowering support teams with smarter tools.Third PrincipleOpenAI in EuropeEU Economic Blueprint14 In manufacturing and automotive,European firms are exploring AI to boost efficiency and crea

82、te smarter products.Mercedes-Benz is integrating ChatGPT into its in-car assistant,aiming to improve the driver experience.More broadly,industrial players across Germany,Italy,and beyond are using AI for predictive maintenance,supply chain optimization,and design keeping Europes industrial base comp

83、etitive.In creative industries,Spotify is partnering with OpenAI to integrate advanced lan-guage models into its platform,enhancing the music streaming experience through personalized recommendations,dynamic playlist generation,and improved search functionalities.In financial services,BBVA is levera

84、ging ChatGPT Enterprise to give its 125,000 employees AI tools that enhance productivity and creativity across departments from credit risk and legal services to customer experience and marketing.These examples demonstrate flagship European brands operating at the cutting edge of AI adoption and rol

85、lout.SMEs&startupsEuropean small and medium-sized enterprises(SMEs)should be encouraged to adopt AI tools to become more efficient and creative.There is already a thriving European start-up ecosystem with a new generation of AI-fueled companies like Synthflow and Pigment using OpenAIs technology to

86、enhance their platforms and ensure that European companies compete globally.However,the AI adoption imperative isnt just about the most agile companies its also about how AI can boost the broader economy.Despite the EUs ambition to become a global leader in AI,adoption rates remain suboptimal.Accord

87、ing to the latest Eurostat statistics from 2024,only 13,5%of European companies are using AI technology.This figure starkly contrasts with the EUs Digital Decade target of achieving 75%AI adoption among businesses by 2030.At a deeper level,a divide exists between larger organizations that demonstrat

88、e higher adoption rate(41%)while SMEs lag behind(11%).Targeted policy interventions are needed to close the adoption gap.Public sectorPublic sector spending accounts for roughly half of EU GDP,with public institutions man-aging vast amounts of data and resources.As such,governments are uniquely posi

89、tioned to lead digital transformation in critical areas such as infrastructure,healthcare,energy,transportation,and a broad range of public services.Currently,AI adoption in the public sector lags behind the private sector.A European Commission study highlighted that complex procurement processes an

90、d concerns about bias in AI decision-making are significant barriers to wider AI implementation in Europes public institutions.Addressing these issues and encouraging broader AI adoption would deliver more effective and efficient public services,saving taxpayers money and freeing resources for incre

91、ased investment in essential areas.OpenAI in EuropeEU Economic Blueprint15To accelerate this transformation,procurement rules embedding specific technological re-quirements should be revised if they hinder AI adoption.Instead,these rules should actively encourage the integration of AI solutions.Gove

92、rnments should also prioritize suppliers and vendors who utilize AI tools effectively.In sectors where the public sector is a primary or significant service provider,governments should remove unnecessary restrictions that prevent sensible and responsible AI use.For example,enabling doctors in public

93、 hospitals to leverage AI could improve patient care and extend healthcare access.Finally,governments should be transparent about AI applications within public services.Clear communication builds trust among citizens and demonstrates the tangible benefits of AI-driven public services.EducationTo dat

94、e,ChatGPT has become a go-to tool for students globally to personalize their education and advance their personal development.Most ChatGPT usersnearly four in fiveare under the age of 35 and the majority of conversations are focused on learning and schoolwork.By embedding AI literacy and usage in cu

95、rricula,ensuring equitable access to AI tools,investing in digital infrastructure and providing professional development for educators,we can ensure students will be better equipped as the workforce of the future.In Estonia,we are proud to work with the Government to provide secure,equitable access

96、to AI across the secondary school system,focused on supporting and empowering teachers.This is part of our effort to support educational institutions across the EU including ESCP France,IE Madrid,University of Pisa,Gothenburg and WHU Germany that have taken steps to make AI as fundamental to their c

97、ampus as using the internet.Educational institutions and governments have a core role in ensuring that all students and teachers have access to AI and receive the necessary training to benefit from its potential.Estonia has fully embraced this challenge and will offer vital learnings to other EU mem

98、ber states.SocietyEuropean governments and industry should work together to scale AI literacy so that users understand these tools and make informed choices.This includes investing in public awareness campaigns and professional training.On a citizen level,libraries and community centers could offer

99、AI familiarization workshops similar to past efforts on basic digital skills.Such initiatives ensure that all Europeans,not just tech experts,have the chance to benefit from AI and use it on their own terms.At OpenAI,we recently announced the next phase of the OpenAI Academy with a publicly availabl

100、e,free online resource hub to support AI literacy and help people from all backgrounds access tools,best practices,and peer insights to use AI more effectively and responsibly.OpenAI in EuropeEU Economic Blueprint16Our policy proposals Introduce fiscal incentives:Tax credits or subsidies for AI-rela

101、ted investments can help de-risk innovation and make adoption more accessible,particularly for SMEs and local authorities with limited resources.Encourage public-private partnerships:Governments should bring together industry,research institutions,and civil society to develop tailored AI solutions a

102、nd support deployment at scale,like the initiative in Estonia.Build a mechanism to share best practices across member states:especially in sectors like manufacturing,health,or public administration to accelerate learning and diffusion.100 Million AI Citizens:Train 100 million Europeans in foundation

103、al AI skills by 2030 through freely accessible online courses in all official EU languages.AI Literacy Ambassadors:Mobilize 10,000 European“AI Literacy Ambassadors”by 2030educators,influencers,and entrepreneursto locally promote and teach AI awareness.Build an AI-native education system:Ensure equit

104、able access to AI in education by embedding AI literacy in secondary and post-secondary curricula,expanding tool access,upgrading digital infrastructure,and training educators.OpenAI in EuropeEU Economic Blueprint17Ensure that AI is built responsibly and reflects European values.The EU and the AI in

105、dustry must work hand-in-hand to ensure that,as AI tools increasingly become part of Europeans daily lives,users can confidently trust that potential harms are being considered and mitigated.Equally important is empowering individuals to meaningfully control and personalize these tools according to

106、their specific needs and preferences.Building this trust will be crucial to ensuring AI is embraced widely and benefits everyone in society.Fourth PrincipleProtecting young peopleYoung people deserve special consideration in the AI regulatory agenda.They are avid users of new technologies and will i

107、nherit the AI-driven world we create.The EU has long recognized the need to protect children and teens online,and AI should be no exception.At OpenAI,weve made youth safety a priority,establishing robust safeguards in our systems and working with trusted organizations dedicated to protecting young p

108、eople.We believe AI companies,together with policymakers,should:Prevent exploitation and harmful content.It is critical to guard against AI being used to generate child sexual abuse material or other exploitative content.We encourage policies and industry standards that prevent the creation or distr

109、ibution of AI-gen-erated Child Sexual Abuse Material(CSAM).This includes building AI systems with safeguards(e.g.,filters,detectors,classifiers)that block attempts to produce such content.OpenAI incorporates protections throughout the AI development life cycle to mitigate misuse.We are collaborating

110、 with organizations and law enforcement to identify and block harmful outputs.The EUs proposed legislation to combat child sexual abuse online can intersect with AI governance,ensuring AI tools do not become a new vector for abuse.Promote safe use of AI.Promote safe use of AI among younger people:AI

111、 systems should prioritize the best interests of young people by upholding their rights and supporting their cognitive and socio-emotional development at every stage.This means designing built-in safeguards,including regularly updated content modera-tion systems that respond to emerging risks and en

112、courage healthy engagement.Transparent practices and appropriate human oversight are essential to ensure AI aligns with young peoples well-being.OpenAI in EuropeEU Economic Blueprint18 Promote AI literacy.Governments and industry should work together to integrate AI education into school curricula,e

113、quipping young people with the skills to under-stand how AI works,its limitations,and how to use it safely and critically.These efforts can build on the EUs Digital Education Action Plan and successful models like our partnership bringing AI to Estonian schools and our support for the Beneficial AI

114、for Children Coalition.Coordinate with child protection stakeholders.Companies like OpenAI should continue to form partnerships with safety organizations globally and in Europe(such as EUROPOLs Cyber Crime Centre or national centers for missing and exploited children),sharing information on emerging

115、 threats and working on better reporting and prevention mechanisms.A cooperative approach between AI providers and law enforcement,under clear legal frameworks,will help anticipate and combat new risks to young people that AI might pose from deepfake abuse imagery to harmful interactions.With youth

116、safety a core component of AI governance,Europe can ensure the next gener-ation grows up with AI systems that support their well-being and development.European parents and educators should have confidence that AI tools whether used for education,entertainment,or social connection come with necessary

117、 protections in place.This is both a moral imperative and key to public trust:if families and educators feel AI is safe for children and teens,they will be more open to its adoption in society at large.Choice and empowermentAI should be a tool for people.That means users whether individual consumers

118、,developers,or businesses should have meaningful control over how they use AI.We think the EU rules should reinforce user agency by allowing people to customize AI tools to their needs.For example,a user of an AI assistant should have settings to determine the assistants style or constraints(within

119、safe bounds).Developers building on AI APIs should similarly have freedom to fine-tune models for their use-cases(again,provided safety standards are met).As AI systems become more adaptive and autonomous,its essential that regulations do not inadvertently hinder beneficial personalization.On the co

120、ntrary,EU policy can play a role in encouraging companies to provide users with meaningful control over how their data is used by AI systemsand how outputs are shaped in return.The ability for individuals to tailor their AI experience enhances both the utility and the societal acceptance of these to

121、ols.OpenAI,for example,enables users to customize elements of ChatGPTs within safe and transparent boundaries,and to decide whether their data is used for model training.Putting users in control is not only a product design choiceits a foundational principle for trustworthy AI.In essence,a user-empo

122、wered approach means AI will be created alongside European citizens,not imposed on them.European consumers should be able to trust that AI services serve their interests,protect their privacy,and exercise choice.OpenAI in EuropeEU Economic Blueprint19Our policy proposals AI Awareness Day:Creation of

123、 an annual moment on a specific aspect of AI (fairness,AI for science,responsible use).The theme will change each year and will drive awareness and conversations in schools and organizations.A Responsible AI Innovation Prize:Inspired by the Sakharov Prize,the European Institutions could award a priz

124、e each year to recognise outstanding,innovative work in AI and related disciplines in Europe.Extend Safer Internet Day to AI:Expand the scope of the EU-supported Safer Internet Day to cover AI,offering resources and guidance tailored to children,parents,and educators.This includes co-developing age-

125、appropriate educational materials,safety guidelines,and awareness campaigns on the responsible and safe use of AI tools.Partner with industry,civil society,and youth organizations to ensure broad reach and relevance.Youth Digital Agency in AI Initiative:Establish an EU-funded initiative that support

126、s the development of AI applications co-designed with young people,particularly those aged 1317.This could involve participatory design sprints with schools,youth councils,and child safety organizations to build features that reflect their needs and values.Voluntary EU Code on Child-Centric AI Desig

127、n and Use:to establish a collaborative,forward-looking framework for AI developers,platforms,civil society,and policymakers to co-develop,adopt and adapt common standards that protect and empower children in the AI ecosystemwhile enabling responsible innovation.OpenAI in EuropeEU Economic Blueprint2

128、0ConclusionThe EUs competitiveness and growth in the AI age depend on embracing innovation and enabling it at scale.By believing in its innovators and giving them the tools and freedom to succeed with sensible guardrails the EU can lead in this new frontier.The payoffs will be substantial:higher pro

129、ductivity growth,new jobs and companies,and solutions to challenges from healthcare to climate that improve citizens lives.By focusing on four pillars compute,energy,data,and talent as well as reducing regulatory burdens,the EU can create an environment where AI innovation flourishes on European soi

130、l.The economic impacts of such investment are numerous:it will create direct jobs(e.g.,constructing a new chip factory or wind farm),indirect jobs(through ecosystems of suppliers and service providers),and induce innovation-led growth(new startups,productivity boosts for companies that adopt AI,etc.

131、).Boosting local economies through these projects also has a regional cohesion effect if done strategically.For exam-ple,AI infrastructure investment can be directed to regions that need economic revital-ization like establishing an AI research center in a city affected by industrial decline,or a da

132、ta center in a rural area with available space and renewable energy potential.Yet time is of the essence.The global demand for compute and energy for AI far outstrips current supply;and as mentioned,large pools of capital(hundreds of billions of euros globally)are searching for places to invest in A

133、I infrastructure.Europe should act fast to channel these resources into projects that support a democratic,sustainable AI ecosystem.If Europe doesnt provide the attractive environment for this investment,through clear policies and possibly public-private partnerships,the funds will flow to projects

134、elsewhere,potentially backed by regimes that do not share Europes values.To put it plainly,infrastructure will define Europes AI destiny.We are optimistic about the EUs AI future.In just two years,our technology has become an indispensable and valued part of millions of European citizens lives.Weve

135、opened our first office in Paris;weve based our regional headquarters in Ireland;weve set up an office in Brussels to work with EU officials;and were now opening a Munich office.We see tremendous promise in the talent and momentum present in Europe and are eager to continue and reinforce our growing

136、 partnerships here.We are ready to do our part whether by collaborating on research,delivering powerful new tools,or working on infrastructure needs to ensure Europe has access to the best AI capabilities.To reflect the rapid evolution of AI technology,this Blueprint is a living document.We will con

137、tinue to update our principles as we learn from working with governments across Europe on how to thoughtfully steward AI technology for the public good.OpenAI strongly believes that with the right vision and execution,Europe can harness AI to solve the grand challenges of our time from improving hea

138、lthcare and education to enabling the green transition and in doing so,deliver wide-spread prosperity and well-being for its citizens.OpenAIs mission is to ensure that as AI advances,it benefits everyone.Were building AI to help people solve hard problems because by helping with the hard problems,AI

139、 can benefit the most people possiblethrough more scientific discoveries,better healthcare and education,and improved productivity.Were off to a strong start,creating freely avail-able intelligence being used by half a billion people around the world,including 3 million developers.We believe AI will scale human ingenuity and drive unprecedented economic growth and new freedoms that help people accomplish what we cant even imagine today.Cover image created with 4o image generationAbout OpenAI