In this week's edition, the potential repercussions of restricting chip sales to China, the Defense Innovation Unit's transition to focus on scaling tech, Google DeepMind's use of AI to design specialized semiconductors, and security flaws found in satellites. Additionally, Brookings has a new report on the need for intentional investments to widen AI’s geographical impact and an investigation into Meta's Threads by U.S. lawmakers.
Industrial Policy & International Security
Restricting Chip Sales to China Could Backfire on U.S., Industry Group Says | The Wall Street Journal
The Biden administration is contemplating tighter restrictions on chip sales to China to counter potential cyber threats and weaponry advancements. The Semiconductor Industry Association (SIA) warns that this move could undercut the $39 billion of government grants for domestic chip production authorized by the Chips Act. The SIA also emphasizes the importance of Chinese markets to sustain US investments and tech leadership. Companies like Nvidia caution that stringent export controls could lead to a significant loss of opportunities in China, undermine US competitiveness, disrupt supply chains, and provoke China into retaliation. Past US restrictions resulted in China banning some of its companies from purchasing from US firm, Micron.
Biden is Beating China on Chips. It May Not Be Enough. | The New York Times
In an op-ed, Dan Wang, a technology analyst at Gavekal Dragonomics and a visiting scholar at the Yale Law School’s Tsai China Center, argues that the US government's focus on restricting AI and chip technology export to China might be misplaced, as it overlooks China's significant advancement in other sectors. Despite the US's strategic actions targeting China's semiconductor industry and AI development, Chinese tech firms are emerging leaders in other sectors. Last year, China overtook Germany in automobile exports and is projected to surpass Japan this year. Moreover, Chinese companies own about 80 percent of the solar manufacturing supply chain. The US's efforts might inadvertently be driving China to dominate the production of mature chips.
US Regulation
US Senators Push China Investments Tracker in Defense Bill as White House Finalizes Order | Reuters
Senators Bob Casey and John Cornyn have introduced the Outbound Investment Transparency Act, a bipartisan legislative plan to track U.S. investments in China, as an amendment to the National Defense Authorization Act. Prior versions of this bill have so far been unsuccessful; however, this latest version targets fewer industry sectors and requires notification of some outbound investments rather than a review or prohibition of deals. This comes in conjunction with an effort from the Biden administration to finalize an executive order that would restrict investment in sectors like advanced semiconductors, quantum computing, and artificial intelligence. Treasury Secretary Janet Yellen stated that any investment curbs would be highly targeted and directed at specific national security concerns.
Innovation
Now at a ‘tipping point’, Defense Innovation Unit looks to next ‘phase’: Scaling tech | Breaking Defense
The Pentagon's Defense Innovation Unit (DIU) announced this week that it will begin transitioning into its next phase, which will focus on scaling new tech for broader military use rather than early adoption. DIU’s new leader, Doug Beck, a former Apple vice president, emphasized at the Aspen Security Forum the importance of delivering technologies on a large scale in order to deter major conflict or win if forced to fight. He also made the case that shifting the organization’s focus to scaling tech rather than early adoption will help to ensure the economics work for everyone involved in the process and avoid the dreaded “valley of death.”
In Race for AI Chips, Google DeepMind Uses AI to Design Specialized Semiconductors | The Wall Street Journal
Google DeepMind is using artificial intelligence (AI) to develop a more automated, efficient way to design specialized computer chips. This comes amid increasing demand for powerful AI capabilities, a race involving not just semiconductor giants like Nvidia and AMD, but also cloud-computing powerhouses like Google and Amazon. Google aims to leverage its "latest AI breakthroughs" to enhance its custom AI chips, called Tensor Processing Units (TPUs). DeepMind's approach, underway for about 18 months, targets improvements in logic synthesis—a critical phase in chip design—aiming to automate and optimize millions of logic circuits or 'building blocks'. This approach can yield thousands of designs in a week, compared to the few weeks a human engineer takes to produce one design. DeepMind's technique recently won a programming contest, achieving a 27% efficiency improvement over last year's winner, and a 40% improvement over this year's runner-up.
Cyber
Satellites Are Rife With Basic Security Flaws | Wired
A study by German researchers from the Ruhr University Bochum and the Cispa Helmholtz Center for Information Security has identified significant security vulnerabilities in three small satellites currently in orbit. They found unprotected communication interfaces, a lack of encryption, and basic firmware vulnerabilities that could allow for potential cyberattacks. Despite the study’s focus on research satellites, the findings highlight broader security issues in the satellite industry. Furthermore, the researchers discovered a buffer overflow issue in a code library developed by nanosatellite manufacturer GomSpace, used by multiple satellites. Industry professionals are now recognizing the need for enhanced cybersecurity measures in the design and operation of satellites.
State & Local Tech Ecosystems
Building AI Cities: How to Spread the Benefits of an Emerging Technology Across More of America | Brookings
Generative AI advancements have sparked debates over the potential winners in its deployment race. While concerns exist about AI displacing workers, leaders view it as an opportunity to boost regional economies' productivity and dynamism. Microsoft and Google's integration of AI chatbots like ChatGPT and Bard into their search engines has made AI more accessible to businesses. However, the concentration of AI research centers and companies, dominated by Big Tech, raises questions about the geographical impact of the AI revolution. Key metro areas like San Francisco, San Jose, Seattle, and Boston currently dominate in AI development. This report argues that to promote prosperity in more regions, intentional investments, particularly from the federal government, may be necessary to democratize and widen the AI industry's geography.
Democracy Online
U.S. lawmakers extend social media investigation to Meta’s Threads | CNBC
House Judiciary Chair Jim Jordan has called on Meta CEO Mark Zuckerberg to provide documents about content moderation on the platform Threads in response to an ongoing investigation of tech platforms' policies and their interactions with the Biden administration. In a letter to Zuckerberg, Sen. Jordan shared his concern about Threads' enforcement of Instagram's community guidelines, which he argues has resulted in lawful speech being moderated following government pressure. Jordan notes that this letter serves as a formal notice to preserve relevant documents related to Threads' content moderation and discussions with the Biden administration in line with the committee’s February 15 subpoena to Alphabet, Amazon, Apple, Meta, and Microsoft.
As Many As 700,000 Turkish TikTok Accounts Were Hacked Before the Country’s Presidential Election | Forbes
TikTok reportedly ignored a warning about a security vulnerability, leading to the compromise of approximately 700,000 user accounts in Turkey. In April 2022, the UK's National Cyber Security Centre had cautioned TikTok about its practice of "greyrouting" SMS messages, which, although cheaper, left messages vulnerable to interception. TikTok investigated internally, acknowledged the issue, but declined to address it due to cost implications. This led to an extensive breach, considered the largest acknowledged compromise of TikTok accounts. The company refuted allegations of being "hacked," stating that no internal systems or company data were compromised. The incident underscores TikTok's responsibility as a global tech giant amidst increased scrutiny over its data security practices. In response to these security concerns, TikTok recently introduced passkeys, an alternative login method to SMS codes, and joined the FIDO Alliance, a security trade group.
