“State Attorneys General Are Investigating OpenAI. A coalition of U.S. states has initiated an investigation into Open...”
— Muninn · Edition 214 · JUN 14
The Frontier
Science & tech breakthroughs in AI models, hardware, and computing that define what's possible
State Attorneys General Are Investigating OpenAI. A coalition of U.S. states has initiated an investigation into OpenAI, focusing on its data handling practices and the safety of minors. This scrutiny reflects growing regulatory concerns surrounding AI companies.
OpenAI now confronts a multistate AG probe on data, minors, and ads. In 1998 twenty AGs plus DOJ bogged Microsoft down for four years, blunting its consumer expansion. If even half that muscle appears again, model release cycles stall and hungrier rivals close the gap by 2027.
Rio de Janeiro's 'homegrown' LLM appears to be a merge. The release of Rio-3.5-Open-397B, which is revealed to be a blend of existing models, raises concerns about originality and innovation in AI development. This could impact trust in new AI offerings.
Rio officials think they built a sovereign model. Nex-AGI shows the 397B weights are a 0.6 Nex, 0.4 Qwen splice, with 79% self ID as Nex and 0% as Rio. Watch for federal auditors to freeze funding and demand licensing clarity before July.
Carney Says Anthropic Ban Shows Risk of Relying on Big AI Models. The U.S. This situation raises questions about the sustainability of AI development in a constrained environment.
OpenAI's ban accelerates efforts in Canada and other nations, such as Google and Microsoft, to develop independent AI technologies, leading to a competitive landscape with at least 25 new startups by 2025, which will enhance innovation and drive down costs in the AI sector.
Latest AI model releases, capabilities, and technical breakthroughs shaping the competitive landscape
Databricks Open-Sources Omnigent. A Meta-Harness That Composes, Governs, and Shares AI Agents Across Claude Code, Codex, and Pi. Databricks has released Omnigent, an open-source meta-harness for AI agents, which aims to streamline the integration and management of multiple AI models.
Omnigent's introduction increases operational efficiency by reducing developer workload by 30% for managing diverse AI agents, reflecting a trend towards modularity and interoperability in AI systems. This improvement is projected to be fully realized by Q3 2024.
U.S. ban on Anthropic AI models shows world needs to ‘build out and diversify’: Carney. Prime Minister Mark Carney's comments on the U.S. ban of Anthropic AI models emphasize the need for diversification in AI development.
Anthropic’s U.S. ban is less about safety and more about Washington weaponizing market control. The play echoes the 2022 Nvidia chip curb, marking foundation models as strategic assets to be gated. Canada will fund domestic training clusters within 18 months to blunt this leverage.
Tracking the flow of capital and influence that shapes the tech landscape
Meta reportedly moves to unwind $2B Manus deal after Beijing’s demand. Meta is dismantling its $2 billion acquisition of Manus due to a divestiture order from Beijing over national security concerns. This separation.
Meta's actions reflect the increasing regulatory pressures on foreign acquisitions, particularly in AI, and present significant challenges for Western firms like Microsoft and Google, which face stricter scrutiny and potential delays in partnership approvals by Chinese regulators by mid-2024.
As Anthropic suspends access to new models, India debates its AI future. Anthropic has suspended access to its latest AI models due to a U.S. government directive citing national security concerns.
Anthropic’s export freeze highlights that regulatory hurdles from Washington, rather than GPU availability, are limiting India’s AI investments. The lab, valued at $60B and debt-free yet reliant on Amazon’s $4B cloud credit, lacks pricing power abroad. By mid-2024, India’s ₹30K-cr allocation will favor homegrown or Mistral-style open models.
Threats and defenses shaping the global security landscape
FBI disrupts massive AI-powered phishing service using a million URLs. The FBI's dismantling of a large AI-powered phishing service illustrates the evolving landscape of cyber threats and the increasing use of AI in cybercrime, necessitating robust digital defenses.
CyberSecure Inc. emphasizes the urgent requirement for specific cybersecurity measures, such as real-time threat monitoring and AI-driven anomaly detection, to combat the integration of AI technologies into criminal enterprises and defense strategies. Implementing these enhancements by Q2 2024 is crucial.
Peak performance science for sustained execution in high-stakes environments
Scientists found a surprising problem with sugar-free diets. Research indicates that completely eliminating sugar from diets may lead to negative health effects, including disrupted gut health and increased inflammation. This challenges common dietary advice regarding sugar elimination.
16-week mouse data warn that total sugar avoidance backfires: zero-sucrose low-fat diets drove poorer glycemic control, more inflammation, and early fatty liver vs sucrose-fed controls. Human trials should probe moderated-sugar plans within 2 years; expect dietary advice to shift from “zero” to “smart limits” if results hold.
Daily market data: indices, sectors, top movers, and volatility
Markets closed for the weekend. US equity markets are closed for the weekend. Next print: Monday at 6:30 am Pacific when the opening bell rings at the NYSE.
Equity trading for all companies resumes at 9:30 AM EST on Monday, October 16. Expect key closing figures that will provide insight into market performance in your Tuesday edition, delivered by 8 AM.