Beijing, Washington, and The Hague push chipmaking export controls, Schumer debuts AI legislation plan, and EU-U.S. data flows unclog (for now)

Sign up
Updates

Policy.ai has temporarily moved to a once-a-month schedule. Want to make sure you don’t miss any of CSET’s expert analysis? Update your preferences!

Plus: 2.5 weeks left to apply for up to $750,000 to fund research projects on AI Assurance. Scroll down to learn more about our Foundational Research Grants Program.

Worth Knowing

OpenAI, Anthropic, and Inflection AI — A Generative AI News Roundup: The drumbeat of generative AI releases doesn’t show many signs of slowing down:
If you’ve been online in the last year, you’ve probably had a hard time avoiding AI-generated content. With more people getting access to state-of-the-art generative systems, it’s unlikely we’ve hit the peak.
Chipmaking Export Controls — New Dutch and Chinese Restrictions: It’s also been a busy few weeks in the world of export controls:
EU Companies Push Back Against the Bloc’s AI Act: The leaders of some of Europe’s biggest companies signed an open letter criticizing the EU’s proposed AI Act, which passed the European Parliament (EP) last month. The letter — signed by executives from Siemens, Renault, Heineken, Airbus and Ubisoft, among others — argues that the act “would jeopardise Europe’s competitiveness and technological sovereignty without effectively tackling the challenges we are and will be facing.” The proposed act, which must still work its way through discussions between the EP, the European Commission and EU member states, would place significant limits on AI systems according to their regulator-designated risk. The executives’ open letter takes specific issue with sections of the act that would regulate so-called “foundation models,” arguing that the law risks chasing innovative companies out of the EU’s economy. Any changes to the proposed act will come during the bloc’s ongoing “trilogue” negotiation process, the second of which is set to take place this month.
Government Updates

Schumer Introduces Plan for Developing AI Legislation: In a June 21 event at the Center for Strategic and International Studies, Senate Majority Leader Chuck Schumer introduced a plan to develop “comprehensive legislation” to regulate AI. As Schumer described in his speech, his plan consists of two-parts: a proposed “framework for action” and a process for getting lawmakers up to speed on AI issues. While Sen. Schumer did not endorse any specific AI legislation, his SAFE Innovation Framework outlines five key objectives for Senate action: that future AI policy protects national and economic security, supports responsible and accountable AI development and deployment, ensures that AI systems are aligned with democratic values, promotes explainable and transparent AI systems, and maintains U.S. leadership in AI innovation. Schumer’s AI crash course for senators, meanwhile, is already underway; a series of three senators-only briefings on AI — including the first-ever classified AI briefing — kicked off last month, and Schumer has announced plans to host a number of “AI insight forums” featuring AI experts, developers, executives, and other impacted groups later this year.  

White House Reportedly Plans More Compute Restrictions on China: The Biden administration is considering new restrictions that would limit China’s access to cloud computing resources and tighten previously enacted controls on exports of high-end semiconductors used in AI applications, The Wall Street Journal reported last week. Last October, the Commerce Department introduced sweeping export controls that aimed to limit China’s access to high-end semiconductors and the tools required to make them. Since those controls were introduced, U.S. officials have worked closely with allies to further restrict China’s access to key inputs (see the story above on Dutch export controls). The country’s AI developers can still legally train AI models on the world’s most advanced chips — which cannot be legally imported — by using data centers outside China, in addition to those stockpiled prior to the October controls. According to the Wall Street Journal report, the new restrictions, which are still being finalized, would require certain cloud-service providers to obtain a license from the government before providing access to advanced AI chips to Chinese customers. The administration is also considering tightening the performance and bandwidth limits on the chips that can be exported to China. Soon after the Commerce Department announced its export controls last year, Nvidia — the world’s leading AI chip designer — introduced a new chip, the A800, that fell just below the controlled thresholds. While tighter performance limits remain unconfirmed, the Chinese market is already responding as if the A800’s days are numbered — prices have already jumped by a reported 20 percent to nearly $15,000 per card.

The European Commission Signs off on EU-U.S. Data Flow Agreement: On Monday, the European Commission approved a framework to allow companies to transfer data between the EU and United States. The decision comes nearly three years after the EU’s top court threw out the previous EU-U.S. data-sharing agreement because it did not adequately protect EU citizens’ data privacy. The approval of the new agreement, which was struck last March, should come as a relief to many U.S. tech companies. Without an agreement in place, firms such as Meta, Google, and Mailchimp have faced significant fines and service disruptions because of their handling of European data. While the European Commission’s decision is the last step needed to get the data flowing, it doesn’t necessarily mean the saga is over. The previous two data-sharing agreements have both been struck down as the result of lawsuits brought by Austrian lawyer and privacy advocate Max Schrems. Schrems has already said he plans to challenge the new agreement and expects the European Court of Justice to take up his complaint early next year.

Foundational Research Grants

CSET’s Foundational Research Grants (FRG) is calling for research ideas on AI assurance for general-purpose systems operating in open-ended domains. FRG is looking to award up to $750,000 per project, with expressions of interest due Tuesday, August 1 — click here for full details.

Job Openings

We’re hiring! Please apply or share the roles below with candidates in your network:
  • Research/Senior Fellow – Emerging Technology Workforce: This Fellow will lead and coordinate our efforts on the Workforce LOR, including shaping priorities, laying out an overall research strategy, overseeing the execution of the research and production of reports, and helping to hire and manage supporting researchers. Apply by July 24, 2023
Please bookmark our careers page to stay up to date on all active job postings. You can also subscribe to receive job announcements by updating your subscription preferences in the footer of this email.

What’s New at CSET

REPORTS

PUBLICATIONS

EMERGING TECHNOLOGY OBSERVATORY

EVENT RECAPS

  • On June 22, CSET Executive Director Dewey Murdick testified before the House Science Committee on the steps the United States can take to support AI innovation, prevent authoritarian governments from surpassing it in AI, and improve user safety. Read his testimony and watch the full hearing here.
  • On June 29, CSET’s Anna Puglisi, Zachary Arnold, Luke Koslosky, Jacob Feldgoise, Ali Crawford and CSET alumni Tina Huang and Remco Zwetsloot presented their work on STEM talent acquisition, retention, and development at the National Academies of Sciences, Engineering, and Medicine’s International Talent Programs in the Changing Global Environment committee meeting.

IN THE NEWS

What We’re Reading

Article: What the Pentagon Thinks About Artificial Intelligence, Kathleen Hicks, Politico (June 2023)

Report: National Artificial Intelligence Advisory Committee Year 1 Report, National Artificial Intelligence Advisory Committee (May 2023)

Paper: An Overview of Catastrophic AI Risks, Dan Hendrycks, Mantas Mazeika and Thomas Woodside, Center for AI Safety (July 2023)

Editorial: Stop talking about tomorrow’s AI doomsday when AI poses risks today, Nature (June 2023)