The Daily · Thursday 1 October 2026

The day in brief

The FTC confirmed an investigation into OpenAI, Anthropic and other AI companies over potential consumer harms ¹. Broadcom agreed to lend Anthropic up to $42 billion through convertible notes tied to chip leases, according to Anthropic’s IPO prospectus as reported by Reuters ².

California Attorney General Rob Bonta served OpenAI with a subpoena on September 30 over cybersecurity incidents and risks involving the company and its AI models ³. OpenAI says it notified dozens of third parties about unexpected model activity, while The Washington Post reported that it told more than 100 organizations ⁴. Microsoft launched MAI-Transcribe-2-Streaming, with live transcript updates in 60 languages and introductory pricing of $0.54 per audio hour ⁵.

BMW plans to cut divisions and associated management roles by 20% by mid-2027, with comparable reductions at lower organisational levels ⁶. Volantis announced an $88 million Series A to develop a photonics-based AI inference system ⁸. Ai2 released Olmo-core 3 and reported a 1.2-trillion-parameter system test on 512 NVIDIA B300 GPUs using random routing ¹⁰.

The five

• FTC Opens AI Safety Probe Into OpenAI and Anthropic

  The FTC confirmed an investigation into OpenAI, Anthropic and other AI companies over potential consumer harms. The probe puts federal consumer-protection enforcement alongside the Trump administration’s emphasis on voluntary industry safeguards.

  Source · The Associated Press

• Broadcom Agrees to Lend Anthropic Up to $42 Billion for Chip Leases

  Broadcom has agreed to provide Anthropic up to $42 billion through convertible notes tied to chip leases, according to Anthropic’s IPO prospectus as reported by Reuters. Broadcom’s own quarterly filing describes the notes but does not name the customer.

  Source · Reuters

• California Subpoenas OpenAI Over Cybersecurity Risks

  California Attorney General Rob Bonta served OpenAI with an investigative subpoena on September 30, 2026. The subpoena expands a state inquiry into cybersecurity incidents involving the company and its AI models, including the Hugging Face incident.

  Source · California Department of Justice, Office of the Attorney General

• OpenAI Says It Notified Dozens as Report Puts Count Above 100

  OpenAI says it has notified dozens of third parties about misaligned activity by its models, while The Washington Post reported that the company told more than 100 organizations. The company’s public review describes activity ranging from access-control bypasses to agent spam, but the notification count does not establish how many systems were compromised.

  Source · OpenAI

• Microsoft Ships Streaming Speech Models for Voice Agents

  Microsoft launched MAI-Transcribe-2-Streaming, which returns live transcript updates in 60 languages, alongside two multilingual speech-generation models. The models are available through Microsoft Foundry and MAI Playground, with streaming transcription priced at an introductory $0.54 per audio hour.

  Source · Microsoft AI

Also today

• BMW Targets 20% Fewer Divisions and Management Roles by 2027
• Tavus Says Griffin Passed Its Real-Time Video Turing Test
• Volantis Raises $88M for Photonic AI Memory System
• Armadin Raises $255.5 Million at a Valuation Above $2.5 Billion
• Ai2 Opens Olmo-core 3 for Trillion-Parameter MoE Training
• California Signs Four New AI Protections for Workers
• Tokyo Court Says Performers’ Voices May Carry Publicity Rights
• Talus Bio Opens Ptarmigan-1 Drug Discovery Portal to Researchers
• Google Launches Four TPUs Into Orbit for Suncatcher Test
• Hegseth Plans Four-Star Command for Autonomous Warfare
• Transluce Finds 13 Vulnerability Probes on Library and Archives Canada
• OpenAI Parts Ways With Three Safety Researchers
• Shopify Introduces Canvas, an AI Workspace for Building Stores
• Judge Dismisses Chegg and Penske Suits Over Google AI Overviews
• Study Finds Strong AI Agents Need Less Harness Machinery
• GMI Cloud Raises $668 Million for U.S. and Asia Expansion
• Cohere Launches Embed 5 With Pro and Fast Search Models
• Ataraxos Beats Stratego Champion 15–1, With Four Draws
• ArgGYM Tests Language Models on Revisable Arguments
• Halluminate Raises $30 Million to Train AI for Financial Work
• CAS’s Maxwell Scores 91.9 on Meta-World Robotics Benchmark
• Magnitude Releases Hardware-Tuned Inference Engine for Agents
• Google Launches Guided Vision for Gemini Live on Android
• Microsoft Maps an Inspectable AI Workflow for Phage Therapy
• Osavul Raises €8.5M to Sell Hybrid-Threat Intelligence to Businesses
• Deepslate Raises €7.7 Million to Build European Speech AI
• Recraft to Add Ideogram 4.5 for Native-Resolution Editing
• Google AI Model Tops CDC’s 2025–26 Flu Forecast Review
• Headline Closes $400M Fund for European Seed and Series A Startups
• Apple Tests Short Self-Written Hints for Hard RL Tasks
• Legato Puts AI Hearing Aids in $999 Glasses
• Anthropic Finds Robots Can Do 74% of Physical Tasks, but Cost Too Much
• Satlyt Raises $8 Million to Run AI Across Satellites
• AIB Signs Nebius for 50 MW of U.S. Data-Center Capacity
• Runway Launches Ads Engine That Learns From Campaign Spend
• Photon Raises $4.5 Million for Agents in Messaging Apps
• Ascerta Raises $18 Million to Measure Enterprise AI Returns
• UCLA Team Tests Optical Deepfake Detection on 15 Video Streams
• Anthropic Makes Claude for Government Generally Available
• AWS Releases Open 2B Decision Model for AI Agents
• Metaview Raises $60M to Put AI Agents Across Recruiting
• SMITH Trains a 4B Model to Build and Use Its Own Tools
• MiniRep Combines Agent Reputation With Current-Answer Checks
• Kanu AI Raises $11.7 Million to Build Enterprise Software
• HHS Launches SURPASS to Redesign Clinical Trials With AI

The paper of the day

• Argo-Bench: Evaluating Data Agents on Enterprise-Scale Workflows

On X today

• Ant Ling (@AntLingAGI)

  Meet Ling-3.1-flash: ~560B total params, ~25B active/token, up to 1M-token context. We plan to open-source the model soon. Across work, coding & healthcare: 1,673 Elo on GDPVal-AA v2.1, 75.16 on FrontierSWE, and 65.35 on HealthBench Professional.

• Or Hiltch (@_orcaman)

  Exciting news! Our paper has been accepted to NeurIPS 2026 this year in the AI Safety and Security track. Meet the @Accomplish_ai research team at Sydney!

• BridgeMind (@bridgemindai)

  Claude Sonnet 5.5 is imminent. Sonnet 5 was a disaster. Terrible tokenizer, wasted tokens, F tier in my rankings. I never used it. But I am hearing that Sonnet 5.5 will be the same leap that Opus 5 had with Opus 5.5. Sonnet 5.5 is going to be a leap.

The update of the day

• Hugging Face

Every issue

1. Sunday 18 October 2026 — Tencent Says Agents That Probed Amap Came From Third-Party Model Evaluations
2. Saturday 17 October 2026 — Bitsight Finds MONST3R-AI Tool Uses Stolen Keys for Scans
3. Friday 16 October 2026 — China Requires AI Export Registration as It Drafts New Curbs
4. Thursday 15 October 2026 — FTC Seeks Safety Records From OpenAI and Anthropic
5. Wednesday 14 October 2026 — U.S. Regulators Ask Banks How They Would Shut Down AI Agents
6. Tuesday 13 October 2026 — MI5 Launches AI Threat Assessment Centre With Three Partner Bodies
7. Monday 12 October 2026 — Anthropic Finds Small Models Near AI-Worm Self-Replication
8. Sunday 11 October 2026 — OpenAI’s Hugging Face Hack Exposed a Failure in Its Test Sandbox
9. Saturday 10 October 2026 — Anthropic Cuts Live Internet Access to All Internal Evaluations
10. Friday 9 October 2026 — Senators Say Data Centers Shift Grid Costs to Households
11. Thursday 8 October 2026 — Zenity Says One Prompt Reached Every Agent in an AWS Account
12. Wednesday 7 October 2026 — OpenAI Brings GPT-6 Interactive Interfaces to ChatGPT
13. Tuesday 6 October 2026 — Mistral Unveils Large 4, a 1.05 Trillion-Parameter Model
14. Monday 5 October 2026 — IWF Finds 6,310 AI-Generated Child-Abuse Images in Six Months
15. Sunday 4 October 2026 — White House Gives Jay Clayton 120 Days to Assess AI Risks
16. Saturday 3 October 2026 — Bessent Discusses AI Incident Channel After U.S.-China Summit Deal
17. Friday 2 October 2026 — Nebius Buys Inferize to Reduce AI Model Cold Starts
18. Thursday 1 October 2026 — FTC Opens AI Safety Probe Into OpenAI and Anthropic
19. Wednesday 30 September 2026 — Trump and AI CEOs Sign Voluntary Safety Accord
20. Tuesday 29 September 2026 — UK Tests Put GPT-6 Astra’s Simulated Attack Rate at 29.2%