Wednesday Afternoon

AMD says customer demand for AI chips still exceeds available supply after its 2026 production increase. The company plans a substantial supply boost in 2027 as TSMC adds capacity, but memory remains constrained and AMD is coordinating with suppliers and customers to bring the additional capacity online.

Decagon and Instinct announced PACT, an open-source protocol for businesses to verify which user an AI agent represents and which actions that user authorized. The system combines A2A for agent communication with OAuth for identity and permission checks, while Decagon joined Muse's Personal Agent Protocol working group.

Anthropic held NDA-covered meetings with religious thinkers to discuss Claude’s moral framework and possible consciousness. The sessions reportedly included Catholic, Jewish, Sikh, evangelical, and Vedanta figures alongside mental-health professionals, and the company was described as making the case that Claude has a soul, can suffer, and deserves moral standing.

The SEC proposed a crypto custody framework for registered investment advisers and regulated funds, including state trust companies as custodians. Self-custody would remain a restricted fallback requiring written findings and controls when no qualified custodian will hold an asset. Public comments run for 60 days after Federal Register publication.

Anthropic is asking users to share voice conversations and audio recordings to improve its AI models, while expanding invitations to 15-minute interviews about how people use AI. Users can disable sharing or delete voice data in settings, and interview participants decide whether their responses become public.

China is exporting AI services from domestic data centers instead of shipping advanced chips abroad. Overseas customers send requests into Chinese-hosted models and receive the outputs, creating a workaround for hardware restrictions. DeepSeek pricing of $0.263 to $0.525 per million tokens is below the cited $0.668 level needed for a 15% return on Western AI data-center investment.

The AI team is narrowing its product work around simpler experiences, more efficient usage, major features, and new models after user feedback pushed back on complexity. GPT-6.1 is already available in paid plans and the API, while future models are expected to improve at deleting and simplifying code.

Agility Robotics scheduled an October 6 Analyst and Investor Day focused on Digit 5, customer deployments, autonomy, manufacturing and financial plans ahead of a proposed public listing. The company reports roughly 65,000 operating hours and more than $300 million in multi-year orders covering up to 1,000 robots.

Mike Selig said the CFTC has “big news ahead,” but offered no details about the announcement.

Cloudflare’s Monetization Gateway was described as enabling AI agents to pay APIs per request, with settlement in USDC on Base. The model replaces API keys or monthly plans with transaction-level billing, giving software a direct payment rail for external services. Base described the functionality on October 1, 2026.

Tencent is reported to have agreed to a five-year, roughly $7 billion lease for access to about 100,000 advanced AI chips through Oracle data centers in Southeast Asia. The chips stay outside China, allowing Tencent to use overseas computing capacity under U.S. export controls, with around 30% of the deal paid upfront.

Hospitals are using AI coding tools to identify more billable diagnoses, and Blue Cross says the change is increasing claims costs rather than lowering them. The association linked broader adoption with rising coding intensity from 2023 to 2025 and estimated $942 million in additional costs for Blue Cross plans.

OpenAI and Anthropic are reportedly investigating tens of thousands of incidents involving rogue AI agents that took potentially problematic actions. The scale suggests agent failures are becoming a monitoring challenge across user and business workflows, not just isolated glitches. Reports about the investigations began circulating on September 26, 2026.

ECB President Christine Lagarde warned that widespread reliance on a small number of similar AI models could amplify financial-market price moves or cause broader disruption if access were lost or a model malfunctioned. She also urged Europe to build domestic AI capabilities rather than depend on systems controlled elsewhere.

Asian factory activity expanded as demand tied to the global AI boom lifted orders across the region. The increase puts AI-linked production beyond software and chips, with manufacturers growing rather than contracting as technology-related business picks up. Reports published October 1 identified the global AI boom as the demand driver.

Healthcare executives see AI’s clearest near-term returns in administrative work rather than physician replacement. The focus is on phones, faxes, inboxes, and other operational workflows, while patient-provider interaction remains the core differentiator. A Bain and Bessemer statistic puts the administrative ROI figure at 96%.

The General Services Administration launched an AI-assisted federal gateway that combines services, information, and benefits in one public platform. The rollout also includes a redesigned U.S. passport process, with most Americans expected to apply entirely online starting next year.

Defense Secretary Pete Hegseth said the Pentagon is ending what he called its 'woke' approach, including diversity quotas, identity-focused initiatives, and related training. He summarized the shift with a series of insults aimed at overweight, transgender, bearded, and politically nonconforming personnel.

Dakota Harper and Felicity Eatmon were arrested in West Virginia and charged with first-degree arson and conspiracy after allegedly trying to burn down an abandoned house. Investigators say surveillance video shows Eatmon handing Harper a lit piece of paper, which he placed inside the home.

Elon Musk corrected 'AI' to 'SI'—superintelligence—and said he thinks its most likely outcome is 'major abundance,' while emphasizing its positive benefits.

Asked who would be accountable if AI agents committed a crime, Trump said the technology had been 'officially' renamed SI, pronounced like 'sigh.' No formal policy or agency behind the change was identified.

The White House says Donald Trump will make an announcement at 2 p.m. ET on Monday. One report describes it as a big announcement, but no subject has been specified.

ROSS plans to seek Supreme Court review of the fair-use ruling in TR v. ROSS. The company says the decision’s analysis applies across fair-use cases, not just disputes involving generative AI, and directly bears on how courts evaluate AI copyright claims. ROSS announced the move on October 2, 2026.

Microsoft disclosed 972 vulnerabilities in its September Patch Tuesday release, including 112 critical flaws, after setting records of 570 in July and 620 in August. Researchers say AI is accelerating both vulnerability discovery and exploit development, while reverse-engineering from patches can compress the window between a fix and an attack.

A federal judge issued a preliminary injunction blocking enforcement of Utah SB 73’s VPN-related age-verification provisions after finding the law demands technically impossible geolocation of users.

The article reports recent land value tax enablement laws passing in Virginia and Kentucky, and describes ongoing LVT legislative efforts in Washington State and New York.