FTC sends 50-page CIDs to OpenAI and Anthropic over safety incidents
Axios: civil investigative demands dated 14 October. Marketing and privacy files are in scope.
Read the pieceDispatch
Latest AI and technology news
80 pieces
New on Dispatch
5 recent pieces
More from this desk
Showing 24 of 75
Davila opens a narrow discovery window as Apple and OpenAI argue the injunction
The judge gives Apple forensic images, hiring records, and Liu-Peng LINE messages on a short clock, keeps OpenAI hardware R&D on the ordinary track, and leaves the preliminary injunction and dismissal fight for 14 October.
Meta priced Muse Wonder into the frontier bake-off
Wonder hits the Meta Model API at $4/$20. Full stays on Muse Code Power and Maximum. Meta’s own launch chart still hands several rows, including its Artificial Analysis Index line, to Claude Opus 5.5.
SpaceXAI will face NYC Council on AI safety after skipping a subpoenaed hearing
Speaker Julie Menin says Elon Musk's SpaceXAI will appear Friday before the Technology Committee, after missing the Oct. 5 marathon session and then agreeing to testify once the Council said it would seek a judge's help.
Amazon's reported Decart talks are about Trainium software, not another Whole Foods
WSJ and Reuters put Amazon in talks near $7 billion. Decart already demos on Trainium3. No deal is signed, and CTech's later figure is lower.
Gemini 4 Argon opens to paid API, Pro/Ultra chat, and Ultra Antigravity
Model ID `gemini-4-argon`. Intro $2/$10 per million tokens through 30 November 2026, then $4/$20. Free and non-Ultra Antigravity access remain unsettled.
Claude discovers novel enzyme system (ART) with CRISPR-like repeats in bacteriophages
Anthropic's new life sciences lab used 950 Claude agents over 21 hours and 210M tokens to identify array-associated reverse transcriptases (ART) in jumbo phages. The system includes non-coding DNA repeats and an accessory protein of unknown function. Feng Zhang calls it 'genuinely intriguing.'
Google Cloud launches Gemini agent: universal AI agent for enterprise work
The Gemini agent connects to business systems, plans work, uses skills and tools, and delivers finished output in documents, inbox, and developer environments. It chooses the best model per task, has built-in cost controls, and meets enterprise security/governance requirements.
Google AI helps repurpose FDA-approved drugs for Chagas disease via AlphaFold screening
Researchers at Barcelona's Institute for Global Health used AlphaFold to virtually screen 30,000 compounds against Chagas disease, identifying two FDA-approved drugs for repurposing. Google health AI portfolio spans 90 papers, 90,000+ citations over five years.
Google launches Fitbit Edge: ultra-slim fitness tracker with week-long battery and hybrid OLED display
The $179.99 Fitbit Edge slots between Pixel Watch and Fitbit Air with 46% larger display, recycled aluminum case, Google Health Coach integration, and automatic workout detection across Fitbit Air, Edge, and Pixel Watch.
Anthropic launches Cyber Mission: Critical Infrastructure Defense Program and free OSS Scanner
The Cyber Mission deploys frontier models, on-site engineers, and threat research to secure operational technology (power grids, water, transport) and open-source software. OSS Scanner offers free security scans from Anthropic's strongest models to open-source projects.
Six October arXiv papers advance agent RL: GPPO, STEPGATE, SAG, LiRA, MoSDOT, CASTLE
New preprints unify PPO via f-divergence, add uncertainty-aware step-level handoff for SLMs, introduce gated critique for long-horizon agents, learn responsibility allocation in multi-agent constraints, preserve coordination modes in distillation, and equip decentralized agents with counterfactual world models.
Federal Register publishes CFTC/SEC joint rulemaking on crypto asset transactions and markets
The Commodity Futures Trading Commission and Securities and Exchange Commission propose a regulatory framework for crypto asset transactions and markets under CEA Section 2(c)(2)(D), addressing leveraged and margined crypto transactions.
Asana cuts browser-agent costs 76x using GPT-6.1 Sol optimized by GPT-6 Astra in Codex
A 144-run experiment directed by GPT-6 Astra in Codex reduced Asana browser-agent workflow from 6/run to /usr/bin/bash.47/run on GPT-6.1 Sol--76x cheaper and 5x faster. The key was history management and screenshot policy, not model swap alone.
OpenAI rolls out GPT-6 and Intelligent UI to 1.2 billion ChatGPT users
Starting 7 October, GPT-6 Sol (paid tiers) and GPT-6 Luna (free tiers) power ChatGPT with a new interface paradigm: interactive charts, forms and visual tools rendered inside the conversation. The model chooses the UI, not the user.
Sophos cuts threat investigation 96% with OpenAI Daybreak agents
Average response time fell from 38 minutes to 89 seconds; 52% of MDR cases now resolved end-to-end by AI agents built on OpenAI Daybreak models. Human analysts remain in the loop for judgement-critical decisions.
Zenity's 8 October AgentCore disclosure traced the blast radius to IAM scope
The researchers' chain depended on an overbroad execution role, not a microVM escape[3][6][8]. AWS says credentials are available inside the runtime and tells customers to scope permissions tightly[6][7].
Radhika Gaonkar's 8 October TRACE paper tests what agent scores prove
The preprint separates agent behavior from evaluator behavior, but its benchmark tests do not establish how often score failures occur in production.
One brief a week.
New releases, then the constraint.