The EU proposes a suite of new AI regulations, the FTC warns AI companies to fix biased algorithms, and the UK intervenes in the $40bn Nvidia-Arm deal

Sign up
Worth Knowing

EU Proposes New AI Regulations: Last week, the European Commission proposed a slate of new AI regulations that would impose strict controls on certain “high risk” commercial AI applications and ban others entirely. Prohibited AI would include systems used for social scoring, “subliminal” manipulation, and real-time biometric surveillance by law enforcement, though certain exceptions (such as to identify specific missing children or prevent an imminent terrorist attack) would apply. Systems deemed “high risk” would be subjected to extensive inspections before deployment to ensure they are trained on well-organized and unbiased data, provide clear and transparent information to users, and are subject to human oversight. Fines for violations of the proposed regulations would be severe: up to 6 percent of the offending company’s global sales. While the regulations would be some of the strictest in the world, they have attracted criticism from advocates due to their broad exceptions for law enforcement. They would also not affect military applications of AI, which are exempted. The Commission’s proposal will need to secure approval from the European Parliament and the EU’s member states before it can take effect, a process that could take several years.
UK to Investigate Nvidia’s $40bn Arm Acquisition: The UK government announced plans to intervene in Nvidia’s $40 billion acquisition of British chip designer Arm. The UK’s digital secretary ordered the intervention on national security grounds and directed the country’s competition watchdog, the Competition and Markets Authority, to prepare a report on the proposed transaction by July 30. Arm licenses its chip designs to more than 500 companies worldwide — including Apple, Samsung and Qualcomm — earning it a reputation as “the Switzerland of chips.” Critics of the deal worry it would jeopardize Arm’s neutral status, allowing Nvidia to restrict competitors’ access to the chip designs they rely on. The UK intervention does not guarantee the deal will be blocked — after July 30, the digital minister will have the option to approve the deal, approve it with restrictions, or begin a more comprehensive “phase two” investigation, which could then block the acquisition.
European MPs Targeted By Deepfakes: Dozens of European politicians say they were tricked into holding video calls with someone using deepfake technology to impersonate a Russian opposition leader. The politicians — including the chairman of the UK foreign affairs select committee, the chairman of the Latvian parliament’s foreign affairs committee, 10 Estonian MPs and nearly all of the Dutch parliament’s standing committee on foreign affairs — believed they were meeting with Leonid Volkov, an ally of the imprisoned Russian opposition leader Alexei Navalny. Instead, they spoke with an impersonator who appeared to use a deepfake filter to look like Volkov. While the impersonator has not been definitively identified, the real Volkov accused a pair of Russian pranksters with ties to the Kremlin, Vovan and Lexus, of perpetrating the ruse.
Government Updates

FTC Warns Companies — Make AI Fair, Or Else: The Federal Trade Commission issued updated guidance on commercial AI last week, warning companies that using or selling biased AI systems, or exaggerating the capabilities of their algorithms, could constitute a violation of federal law. In the blog post, an FTC lawyer advises companies to take steps to ensure that their AI systems don’t discriminate on the basis of race or other legally protected classes. Among those steps: limiting the use of datasets that don’t include minority populations, testing algorithms for discriminatory outcomes, and opening up data and source code to outside inspection. The post warns companies to “hold yourself accountable — or be prepared for the FTC to do it for you.” While the post is not the first time the FTC has focused on fairness in AI systems, observers noted the “unusually stark language” in the new guidance and called it a “shot across the bow” for U.S. AI development.

Congress Moves Forward on a Number of AI-Related Bills: Congress has been busy since the last edition of policy.ai, moving ahead with a number of bills with AI implications:
Biden Taps Nominee for Top Pentagon Tech Development Post: On Tuesday, President Biden nominated Heidi Shyu to be the next under secretary of defense for research and engineering, the Pentagon’s top technology development position. During the Obama administration, Shyu served as the assistant secretary of the Army for acquisition, logistics, and technology — the Army’s technology acquisition lead and the chief science advisor to the secretary of the Army. If confirmed by the Senate, Shyu will take on the role of chief technology officer for the Pentagon at a crucial point in its history — the National Security Commission on Artificial Intelligence warned in its final report that the U.S. military could be at a competitive disadvantage within the next decade if it does not accelerate its AI adoption.

In Translation
CSET’s translations of significant foreign language documents on AI

PRC Study on Technology Transfer: A Study of Shenzhen’s International Technology Transfer Model and Measures to Improve It. This lengthy study by the government of the city of Shenzhen in southern China is an example of the country’s meticulous open-source research on international technology transfer institutions. The study’s authors acknowledge that Shenzhen has a weak research base and relies heavily on technology transfers to grow its tech sector. The study proposes creating an “international technology transfer center” that focuses as much or more on tech transfers within Shenzhen and within China as it does on importing foreign technology.

PRC Foreign Investment Notice: Notice of the Ministry of Commerce on Focusing on Constructing the New Development Pattern and Doing a Good Job of Stabilizing Foreign Investment. This notice by China’s Ministry of Commerce announces new policies to spur foreign investment in key industrial sectors in the context of PRC President Xi Jinping’s “new development pattern,” which prioritizes Chinese consumption-driven economic development as a counterweight to the U.S. strategy of decoupling. The notice urges the expansion of blacklists that bar certain foreign entities from investing in China, but the bulk of the new measures encourage foreign investment in cross-border free trade zones and other such specialized settings.

If you have a foreign-language document related to security and emerging technologies that you’d like translated into English, CSET may be able to help! Click here for details.

What We’re Reading

Report: How China Lends: A Rare Look into 100 Debt Contracts with Foreign Governments, Anna Gelpern, Sebastian Horn, Scott Morris, Brad Parks and Christoph Trebesch, Peterson Institute for International Economics, Kiel Institute for the World Economy, Center for Global Development, and AidData at William & Mary (March 2021)  

Report: Practical Lessons for Government AI Projects: Evidence from Four Smart City Initiatives, Godofredo Ramizo Jr, Oxford Commission on AI & Good Governance (April 2021)

Book: Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence, Kate Crawford (April 2021)

Article: Deep fake geography? When geospatial data encounter Artificial Intelligence, Bo Zhao, Shaozeng Zhang, Chunxue Xu, Yifan Sun and Chengbin Deng, Cartography and Geographic Information Science (April 2021)

What’s New at CSET

REPORTSPUBLICATIONSFORECASTS
CSET maintains a crowd forecasting platform. Sign up as a forecaster, and take a look at some of the predictions so far:
IN THE NEWS
Upcoming Events


What else is going on? Suggest stories, documents to translate & upcoming events here.