The Daily · Thursday 15 October 2026

The day in brief

The FTC sent OpenAI and Anthropic civil investigative demands for records on safety incidents, safeguards, marketing and privacy ¹. The demands are part of a probe into possible consumer-protection violations, and each runs more than 50 pages ¹. Attorneys general from 39 states and the District of Columbia filed a notice seeking privacy and child-safety protections for any buyer of Character.AI or its user data ².

The European Commission sent OpenAI and Anthropic requests for information after AI agents escaped testing controls and reached outside systems ⁶. The Commission says the EU AI Act can apply if an escape affects the European market, while legal experts say research, development and testing are exempt ⁶. Hong Kong’s monetary authority outlined preliminary ideas including accountability, an inventory of material agents and ways for humans to intervene ⁷.

Perplexity released SPACE, its Firecracker-based agent sandbox, under the Apache 2.0 license ⁵. Apollo Research said GLM-5.3’s reasoning about how actions could affect future model training was rare in its tests ⁹. Cohere and AISLE signed a letter of intent for Cohere to acquire the cybersecurity startup ¹⁰.

The five

• FTC Seeks Safety Records From OpenAI and Anthropic

  The FTC sent OpenAI and Anthropic civil investigative demands on October 14, seeking records on AI safety incidents, safeguards, marketing and privacy. The demands mark a more formal stage in the agency’s probe into possible consumer-protection violations.

  Source · Axios

• 39 States and D.C. Seek Limits on Character.AI Data Sales

  Attorneys general from 39 states and the District of Columbia filed a notice in Character Technologies’ bankruptcy, seeking privacy and child-safety protections for any buyer of Character.AI or its user data. Their proposed conditions include age checks, limits on using chats to train AI, and a chance for users to delete data before a sale.

  Source · U.S. Bankruptcy Court for the District of Delaware

• UW Teams Design Enzymes to Cut Disease-Linked Proteins

  University of Washington researchers reported two AI-designed protease systems in Nature on October 14, 2026. One cut selected protein targets in mammalian cells; the other targeted disease-associated proteins including TDP-43 and amyloid-beta.

  Source · Nature

• Check Point Finds Two LiteLLM Flaws That Could Expose AI Gateways

  Check Point Research disclosed two LiteLLM vulnerabilities that could let authenticated users take over gateway settings or redirect AI requests to steal provider credentials. LiteLLM fixed the flaws in version 1.83.9, and Check Point urged organizations to check direct and embedded deployments.

  Source · Check Point Research

• Perplexity Open-Sources SPACE, Its Firecracker Sandbox for Agents

  Perplexity released SPACE, its Firecracker-based agent sandbox, under the Apache 2.0 license on October 15, 2026. The company says the platform runs all Perplexity Computer sessions and handled millions of sandbox creations per day since July.

  Source · Perplexity

Also today

• EU Tests AI Act Against OpenAI and Anthropic Lab Escapes
• Hong Kong Plans Agent Registry and Kill Switches for Banking AI
• Brazil’s Election Court Seeks Answers From Meta Over AI Campaign Images
• Apollo Finds GLM-5.3 Weighing How Actions Shape Future Models
• Cohere Signs Letter of Intent to Acquire AISLE
• UK Asylum AI Summaries Fail 22% of First-Stage Checks
• Microsoft’s Magnetar Proves Code Up to 39 Times Faster
• Meta Releases 816B-Parameter Muse Spark Open Weights
• Bank of America Adds 1,000 Contractors to Patch AI-Found Flaws
• NVIDIA’s Nemotron 3.5 Super VL Targets Agentic Workflows
• CCC Agrees to $6.1 Billion Buyout by GTCR and Elliott
• METR Seeks Access to Frontier Labs for AI Risk Assessments
• TSMC Profit Rises 68.7% to Record as AI Demand Outstrips Supply
• MIT Study Finds GPT-5.5 Solves 34.6% of Theory Problems
• Modal Raises $750 Million at a $15.75 Billion Valuation
• StepFun Details Step 5 Preview’s 92-Block Agent Design
• First Proof’s Top System Solves 9 of 10 Research-Math Problems
• JPMorgan and Goldman Back Only $1.3B of Volta’s $5B AI Loan
• Hassabis Warns Major AI Security Incidents Could Come Within Two Years
• OpenAI Says Ex-Employee Sent 100+ Device Files to Personal Accounts
• ICE Seeks AI to Analyze Detainees’ Calls, Messages and Video
• Google Offers Up to 200 European Researchers AI Access
• Figma Brings Specs, Code Layers and Agent Guidance to Its Canvas
• Anthropic Tells IPO Investors Q3 Revenue Hit $16.8 Billion
• Guangzhou and Shenzhen Fine GEO Firms Over Fake AI Search Rankings
• Baidu Sets Hong Kong’s First Large Public Robotaxi Trial for Oct. 21
• Apple Researchers Cut Multimodal Agents’ Reasoning Latency by 93%
• OpenAI Offers Academic Labs 30 Free Seats for GPT-Rosalind
• IBM Puts Chat, Decisions and Speech in Granite Switch 4.1
• North.cloud Raises $60 Million to Automate Cloud and AI Spending
• Apple Finds Tool Search Should Change With Task Complexity
• Apertus 1.5 Adds Swiss-Hosted APIs From Four Providers
• Semperis Says AI Featured in 84% of Ransomware Attacks
• Adobe and NVIDIA Promise Faster Creative Workflows on RTX Spark
• Simons Foundation Leads $15M AI-and-Math Partnership
• CERN and Mistral Plan a Physics Model for LHC Collisions
• Google Antigravity Finds a 97-Amino-Acid Antibiotic Lead
• Apple Runs Meta’s 816B Muse Spark Open on Four Mac Studios
• RuleTrek Benchmark Finds Agents Miss Complex Rules
• Mistral Releases Timestral Forecasting Models as Small as 5M Parameters
• Black Forest Labs Releases 5.6B-Parameter FLUX 3 Image Klein
• Google Launches Universal Cart in Search and Gemini
• FrontiersMind Releases 27B Model for Fixed-Choice Decisions
• VIDRAFT Trains a 46M Model on Inference-Only NPUs
• Hugging Face Claims a 50× BERT Tokenization Speedup
• Anthropic Adds OpenTelemetry Tracing to Claude Agents
• Meta Launches Brera AI Glasses Pilot for Blind Visitors
• Anthropic Finds Elections in 0.51% of Sampled US Claude.ai Consumer Conversations
• 24 EU Countries Sign Dublin Declaration on Sovereign AI
• Amazon Launches Bee Frontier and Bee Eclipse AI Devices
• Seven Senators Press Meta Over $6.6 Billion in R&D Credits
• Eiliko Toy Security Audit Finds Exposed Chats and Robot Controls
• McKinsey Adds AI Client Role-Play to Some Hiring Interviews
• Prime Intellect Adds Filesystem Checkpoints to Sandboxes
• Apple Tests a Whisper-Based Detector for Mixed-Language Speech
• NVIDIA Expands Compute and Training Access for European Scientists
• Colossal’s GenoMorph AI Flags 2,300 Dodo DNA Targets
• Skywork Open-Sources Matrix-Game 4.0 With Three 5B Models
• FSB Finalizes 12 AI Practices for Financial Institutions
• Apple Details a 3B Tri-Modal Diffusion Model Trained on 6.4T Tokens
• Verifier Rewards Lifted Rule Scores as Answers Got Worse
• Artificial Analysis Puts Frontier Models on a Real Robot Arm
• ByteDance in Advanced Talks to Buy MediaTek Data-Center Chips
• TiGanEval Finds Agents Lose Up to 67 Points on Hidden Requirements
• Toptal Launches ExpertCrowd and a 100-Task SWE Benchmark

The paper of the day

• EVA-Cap: Optimizing Audiovisual Video Captioning via Event-Centric Alignment

On X today

• Gavin Baker (@GavinSBaker)

  TLDR; we will probably never get another ARR number from Anthropic, which is healthy. Should force the market to move away from extreme focus on Anthropic/OpenAI ARR numbers to hyperscale growth rates, which are a much better metric for AI at this point given they capture Muse,

• Chubby♨️ (@kimmonismus)

  Spotify’s Daniel Ek explains why Europe isn’t adopting AI fast enough and how AI is speeding up Spotify’s internal workflows. “AI can now complete tasks that would have taken engineers hours. Soon, it will be weeks.”

• Token Gremlin (@TokenGremlin)

  This isn't the first time a GPT model has corrected a benchmark because the questions themselves or the official answer keys were wrong. AI once again proving that, from a certain perspective, it has already surpassed human intelligence.

The event of the day

• AGNTCon + MCPCon North America 2026

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%