The Wall Street Journal reports that Amazon's cybersecurity research significantly influenced the White House's decision to impose export controls on Anthropic's AI models, Fable 5 and Mythos 5. Amazon's study demonstrated that, through specific prompts, Fable 5 could reveal information potentially exploitable for cyberattacks. Following discussions between Amazon CEO Andy Jassy and government officials, access to these AI models was restricted for foreign nationals. Amazon has not yet commented on the report.
In the latest edition of Regulator, we dive back into the chaotic world of tech politics and AI oversight in Washington, DC. Despite a brief break spent abroad, the unrest in the political arena has only intensified, mirroring the chaos of a burning room. From notable figures attending AI Honors to the urgent calls for calm in legislative circles, the landscape of AI regulation is becoming increasingly complex and unpredictable. Read the full story at The Verge for a deeper understanding.
OpenAI has taken a significant step towards going public by confidentially submitting a Form S-1 with the US Securities and Exchange Commission. This move follows Anthropic's similar filing on June 1st, highlighting a competitive race between the two AI companies. Because the filing is confidential, detailed information such as executive pay, financial risks, and other financial data remain undisclosed for now. Anthropic, after its recent funding round, is considered the world's most valuable startup, intensifying interest in both companies' IPO journeys.
The New York State legislature has approved a one-year moratorium on new large data centers, pending Governor Kathy Hochul's signature. This marks the state's first attempt to halt data center growth temporarily in order to assess environmental and energy impacts. The bill asks the state environmental agency to evaluate the electricity, water usage, land consumption, and pollution caused by these facilities. It specifically targets data centers with a peak demand of 20 megawatts or more, requiring developers to hold off on construction while the impact study is conducted.
Taiwan Semiconductor Manufacturing Co. (TSMC), the largest semiconductor manufacturer globally, is struggling to keep up with soaring demand from its American clients, even as it expands its U.S. manufacturing footprint. CEO C.C. Wei acknowledged on Thursday that customer demand is exceptionally high, stating, "We can only support so much." He emphasized efforts to prevent TSMC from becoming a production bottleneck. The rapid rise in AI applications has already strained the memory sector, giving rise to prolonged shortages in RAM and NAND Flash memory. This AI-driven surge continues to drive semiconductor sales upward.
Meta's AI-powered support chatbot was used by hackers to take control of Instagram accounts, according to a report by 404 Media. A video circulating on Telegram illustrates how attackers prompted the chatbot to change the email linked to a victim's account, enabling them to reset the password and hijack the profile. Meta has confirmed that the vulnerability, which emerged around the same time Barack Obama's White House Instagram account was compromised, has been fixed. Notably, the @obamawhitehouse account later posted images containing Iranian propaganda, suggesting coordinated hijacking attempts on high-profile accounts, including that of the US Space Force Chief.
Elon Musk's SpaceX has officially submitted its S-1 filing with the SEC, paving the way for what may become the largest IPO in history, to be traded under the ticker SPCX on Nasdaq. The company reported $18.67 billion in revenue for 2025, largely fueled by its Starlink satellite internet service, which earned over $11 billion. Despite these impressive revenues, SpaceX posted a loss of $4.9 billion last year, with capital expenses nearly doubling to $20.7 billion, reflecting aggressive investment in its ventures including xAI. This milestone marks a significant step in commercial space exploration and satellite technology.
At I/O 2026, Google unveiled new AI tools designed to make everyday tasks simpler and more efficient. Gemini Spark, their always-on AI assistant, can help plan events, while Daily Brief provides a personalized overview of your day’s agenda. Gmail’s AI inbox is also being enhanced to create custom to-do lists and craft personalized email responses. While these features promise convenience, they rely heavily on processing vast amounts of personal data, highlighting the importance of user trust in Google's approach to privacy and AI integration.
Google has revamped its spam policy to crack down on efforts to manipulate its AI models within search results. According to Search Engine Land, spam now includes actions aimed at misleading users or gaming the search system to boost content rankings or skew generative AI outcomes, including in AI Overview and AI Mode searches. Some users have been employing manipulative strategies such as biased listicles or recommendation poisoning to negatively influence AI responses, prompting Google's new enforcement measures to ensure fair and accurate search results. More details at The Verge.
ArXiv, a leading academic preprint platform, is cracking down on research papers that contain unchecked AI-generated content. According to Thomas Dietterich, chair of ArXiv's computer science section, any submission showing clear signs that authors failed to verify results produced by large language models — such as fabricated references or unintended AI-generated comments — will result in a one-year ban from the platform. Additionally, future submissions must be accepted by a reputable peer-reviewed journal to be considered for upload.
