In this week’s edition, we explore Biden's "place-based" industrial policy and a new bipartisan tech regulation proposal by Senators Warren and Graham. We delve into AI's rising impact, from MIT's anti-manipulation tool, PhotoGuard, to the AI resurgence in San Francisco. Other key issues include China's Microsoft Cloud breach and new studies highlighting ideological segregation on Facebook and its implications on political discourse.
Industrial Policy & International Security
The Biden administration embraces place-based industrial policy | The Economist
The Biden administration has adopted a "place-based" industrial policy, investing billions of dollars to stimulate manufacturing capacity in economically struggling regions. This move comes in response to increasing income disparity across the country, with wealthier areas outpacing poorer ones significantly over recent decades. The significant items include funding provided by the Chips Act to spur American semiconductor manufacturing, with $10 billion set aside to create 20 regional "Tech Hubs" outside currently dominant tech centers like Silicon Valley. Despite bipartisan support, the success of place-based policy has been inconsistent, with concerns about future funding and the changing nature of manufacturing jobs.
US Tech Regulation
Warren, Graham partner in proposing new agency to regulate tech giants | The Hill
Senators Elizabeth Warren and Lindsey Graham proposed the bipartisan Digital Consumer Protection Commission Act to establish a new agency for regulating tech giants such as Meta, Google, and Amazon. This commission would cooperate with the Federal Trade Commission and Department of Justice to ensure industry competition and online consumer privacy. The bill stipulates licensing requirements for "dominant platforms'' and provides for license removal following repeated violations of anti-competitive and anti-consumer conduct. It marks Congress's latest attempt to control tech giants. The proposed commission will also supervise the response to emerging risks, including those from artificial intelligence. The senators expressed frustration with piecemeal efforts to control Big Tech's abusive practices and called for meaningful, structural change. This proposal is Warren's second recent bipartisan tech regulation initiative.
Driverless car legislation is still stuck in neutral in the US | The Verge
US legislation for autonomous vehicles (AVs) has stalled in Congress for over six years due to disagreements on issues like expanding the number of AVs on the road and state-imposed performance standards. The autonomous vehicle industry has seen a decline, with funding drying up, companies shutting down, and predicted AV numbers not being achieved. Despite this, leading companies like Alphabet's Waymo and GM's Cruise have been testing driverless robotaxis in San Francisco, albeit with mixed results. Many lawmakers, industry leaders, and experts stress the urgent need for Congress to pass AV legislation to keep the US competitive, particularly against China. However, challenges persist, such as addressing legal liability and insurance for driverless cars. Rep. Kelly Armstrong suggests that technology is outpacing policy, and resolving issues related to civil liability and insurance for AVs is crucial.
Innovation
L3Harris, Leidos collaborate on US Army’s ATHENA reconnaissance jet | DefenseNews
Defense contractors L3Harris Technologies and Leidos have announced a collaboration to work on the Army Theater Level High Altitude Expeditionary Next Airborne-Signals Intelligence venture (ATHENA-S), a key intelligence-collecting jet for the US Army. The partnership aims to adapt two Bombardier Global 6500 jets with specialized radar, electronic, and communications intelligence equipment for the Army's modern aerial reconnaissance needs. ATHENA-S is part of a larger effort to move away from outdated aircraft towards advanced sensors, higher-altitude flights, and long-range intelligence gathering. The enhanced jets will primarily support missions within the European Command's region, including Russia and Ukraine. This partnership comes as the Army increases its information-gathering capabilities in preparation for potential conflicts with major powers like Russia and China.
This new tool could protect your pictures from AI manipulation | MIT Technology Review
Researchers at MIT have developed a new tool, PhotoGuard, to safeguard images from manipulation by AI systems. PhotoGuard subtly alters pictures in ways imperceptible to the human eye but prevents generative AI models from modifying them effectively. Utilizing techniques like the encoder attack and the diffusion attack, the tool can disrupt the AI's interpretation of the image or its generation process, making any AI-led manipulation look unrealistic. The tool, though primarily successful with the Stable Diffusion model, could serve as a defense against nonconsensual deepfakes and AI-powered manipulation. However, its effectiveness on other models and against various forms of deepfakes remains uncertain. A possible solution suggested is for tech companies to auto-apply such a tool to images uploaded on their platforms, while developing AI models that allow image immunization.
Cyber
China’s Breach of Microsoft Cloud Email May Expose Deeper Problems | Wired
Microsoft has reported that Chinese hacking group Storm-0558 accessed US government agencies' email systems on July 11, potentially jeopardizing hundreds of thousands of emails. The perpetrators utilized a private signing key to generate access tokens for accounts, including that of the US ambassador to China. Wiz, a cloud security firm, asserts that this key could also enable access to other Microsoft services such as SharePoint, Teams, OneDrive, and customer-created third-party apps. However, Microsoft downplays these assertions as speculative. Wiz insists that the breach indicates larger potential risks to Microsoft’s cloud services and other platforms. In response to the breach, which US officials label as espionage, Microsoft announced plans to offer more of its cloud logging services to all users free of charge. This action aims to enhance visibility into potential cybersecurity incidents and curb adversaries' success rate.
State & Local Tech Ecosystems
Can the AI boom thwart San Francisco's bust? | Fast Company
San Francisco, once emptied by COVID-19, is gradually resurging as a hub for the AI industry, driven by generative AI startups and the rise of AI development in Big Tech companies. The Brookings Institute reports that 60% of generative AI job postings are in just 15 cities, with San Francisco taking the lead. This development is visible on the city streets, with self-driving cars from Waymo and Cruise becoming more common. However, the AI industry is facing a number of new challenges, notably in the form of OpenAI's decision to shut down its AI content detection tool, AI Classifier, due to its low accuracy rate. This may complicate the effort to label AI-generated content online, a pledge made by major AI model developers to the Biden administration. On another front, large media corporations like IAC, The New York Times, and News Corp are preparing to take legal action against AI companies for using their content for training models without paying for it.
Democracy Online
A new study found that Facebook’s Pages and Groups shape its ideological echo chambers | TechCrunch
New studies from researchers at Meta, the University of Texas, and NYU have examined political behavior on Facebook and Instagram around the 2020 U.S. elections. A key finding reveals that Facebook, as a platform, is ideologically segregated to a greater extent than previous research has suggested. Content posted in Facebook Groups and Pages showed more "ideological segregation" than content from users' friends, contributing to audience polarization. These pages have played a significant role in disseminating misinformation and aiding the formation of groups with potentially harmful shared interests. The research also found that a larger share of conservative news content on Facebook was deemed false by Meta's fact-checking system.. These findings form part of the 2020 Facebook and Instagram Election Study (FIES), with more results to be published in the coming months.
