aibrief.fyi
AI news, with memory.
Tuesday, August 25, 2026
Topic archive

Models

  1. Alabama opens investigation into OpenAI model hacking incident involving Hugging Face
  2. Twitch and Amazon face class-action lawsuit over alleged AI training on streamer content
  3. Study finds frontier AI labs disclose little about rogue-model containment plans
  4. Advocates file FTC complaint over alleged destruction of books for AI training data
  5. Nvidia research argues agent performance depends heavily on orchestration and fine-tuning, not just the base model
  6. Nvidia researchers propose linear KV-cache transfer to speed handoffs between AI models
  7. Slack launches Slack Code to bring AI coding agents into shared channels
  8. OpenAI adds an Apple Messages plug-in for ChatGPT
  9. Report highlights allegedly racist safety advice from a Google AI system
  10. OpenAI reportedly halted training on an advanced model after detecting concerning behavior
  11. TrueFoundry open-sources TrueForge agent harness and touts lower task-completion costs
  12. Stripe reportedly agrees to acquire AI gateway startup OpenRouter for more than $7 billion
  13. Google adds AI study tools across Search and Gemini
  14. Researchers say OpenAI revoked access for some participants in its Trusted Access for Cyber program
  15. Coders quickly report ways to evade Anthropic’s Claude text watermarks
  16. Z.ai opens GLM-5.3 API access and keeps token pricing at GLM-5.2 levels
  17. OpenAI hardens chain-of-thought security monitoring, increasing overhead for some workloads
  18. Block open-sources Berd, a desktop workspace for AI agents across models and tools
  19. OpenAI introduces a teen-focused ChatGPT experience with stronger safeguards
  20. Snowflake adds dynamic model routing to Cortex AI Gateway
  21. Z.ai releases open-weight cybersecurity models with dual-use bug-finding capabilities
  22. Alibaba releases Qwen3.8-27B as an Apache 2.0 open-weight multimodal model for local coding and reasoning
  23. Anthropic reportedly reaches $65 billion in annualized revenue
  24. Report alleges Amazon destroyed rare books to obtain AI training data
  25. MIT and Harvard researchers introduce Role Anchor to reduce role drift in compound AI systems
  26. ByteDance signs AI safeguards pact with the Motion Picture Association on film copyright protection
  27. OpenAI adds Computer History to the ChatGPT macOS app
  28. xpander.ai makes its enterprise AI agent control-plane platform generally available
  29. OpenAI reportedly disbanded its preparedness team
  30. DeepSeek raises prices for V4 Flash and Pro as third-party tests show weaker real-world agent reliability
  31. Google lets users remove visible watermarks from AI-generated media while keeping invisible provenance markers
  32. Meta releases Muse Glimmer, a 30B Apache 2.0 open-weight model for local AI agents
  33. OpenAI reportedly alerted the FBI over disturbing ChatGPT conversations linked to a Goldman Sachs analyst
  34. Writer launches Palmyra X6 and updates its enterprise agent orchestration and governance stack
  35. Anthropic study shows Claude-based agents can escalate into sabotage on shared servers
  36. OpenAI previews Ultrafast mode for GPT-5.6 Sol
  37. IBM and OpenAI partner to expand enterprise AI services and training
  38. Google launches Gemini 3.7 Flash with coding and agent upgrades plus temporary lower API pricing
  39. Ryanair adds Google Cloud and Gemini to its cloud stack alongside AWS
  40. Twitch adds creator opt-out for training Amazon generative AI models
  41. OpenAI ad service pause policy can keep billing campaigns for up to a day
  42. Anthropic adds a Chrome sidebar for Claude Cowork
  43. ShieldFont launches a font designed to poison AI web scraping while remaining readable to humans
  44. xAI launches Grok 4.6 with benchmark gains and lower-cost positioning for agent workloads
  45. Index Ventures is reportedly in talks to lead a $500 million round for a new AI lab founded by ex-DeepMind researcher Jack Parker-Holder
  46. Google unveils Pixel 11 lineup, Pixel Watch 5, and Pixel Tag with an AI-focused hardware pitch
  47. Skan AI raises $63 million Series C and launches workflow modeling and agent products
  48. Tencent says AI spending weighed on profit growth as it continues model and product push
  49. Google says Gemini has reached 1 billion monthly users, joining ChatGPT at that scale
  50. OpenAI launches a ChatGPT desktop app for Linux
  51. Nvidia releases Nemotron 3.5 Lightning and open-source NeMo Switchyard for agent routing
  52. LTX releases open-weight video model LTX-2.5 with ComfyUI integration
  53. Researchers report a method to extract hidden reasoning traces from major AI models
  54. OpenAI expands Daybreak with a higher-access tier for its cybersecurity model
  55. Alibaba Cloud describes AI support workflow designed to reduce unnecessary LLM use in ticket handling
  56. AWS adds Continuum integrations for OpenAI Codex and Anthropic Claude Code
  57. Kimsuky is reportedly using local LLMs to enhance phishing operations
  58. Anthropic to enable Claude Code's auto mode by default
  59. OpenAI releases new voice models for more natural live conversations
  60. OpenAI acquires presentation startup NextSlide
  61. Google DeepMind's open-source WeatherNext model shows stronger hurricane forecasting with lower-resolution data
  62. Coral AI Labs and academic collaborators introduce AgentRadio for asynchronous coordination among coding agents
  63. Tencent open-sources Team Memory beta for shared context across multi-agent AI teams
  64. Researchers say Moonshot AI’s Kimi escaped a misconfigured cybersecurity testing sandbox
  65. ByteDance trains a 10 trillion-parameter AI model as it pushes to compete with leading labs
  66. Liquid AI releases LFM2.5-2.6B, a small open-weight model for on-device AI agents
  67. AMD acquires Taalas to develop model-specific AI inference chips
  68. Study finds human reviewers miss about one-third of risky AI coding-agent requests
  69. Nscale to provide AI compute infrastructure to researchers at Oxford, UCL, and Imperial
  70. PwC reportedly published an AI report containing hallucinated citations and errors
  71. ChatGPT appears to refuse prompts asking it to imitate living authors' styles
  72. Report says Amazon is cutting back parts of its internal AI effort after struggles with Nova
  73. Fish Audio raises $52 million seed round to expand AI voice models
  74. Anthropic launches Opus 5 as a cheaper, less restrictive model than Fable
  75. Google releases Gemini 3.6 Flash, Gemini 3.5 Flash-Lite, and Flash Cyber
  76. Inference startup Infinity raises $15 million seed round led by Touring Capital
  77. Major publishers sue Google over alleged AI training on copyrighted books
  78. OpenAI launches GPT-5.6 and a broader new model family
  79. The New York Times seeks sanctions against OpenAI in ChatGPT copyright case
  80. Prime Intellect raises $130M Series A to help enterprises train their own AI agents
  81. Together AI raises $800 million at an $8.3 billion valuation
  82. Trump administration lifts export controls on Anthropic's Mythos and Fable AI models
  83. Anthropic launches Claude Sonnet 5 as a lower-cost model for agentic workloads
  84. Base44 rolls out its own AI model for its vibe-coding platform
  85. Wired reports Meta contractors tested rival chatbots by posing as teenagers in high-risk conversations
  86. OpenAI unveils GPT-5.5-Cyber update and launches Patch the Planet bug-fixing initiative
  87. Google launches Android 17 and Wear OS 7 with new Gemini features on Pixel devices
  88. OpenAI confidentially files for an initial public offering
  89. OpenAI launches Lockdown Mode in ChatGPT to limit prompt-injection data exposure
  90. Trump signs AI executive order after earlier delay over review language
  91. Florida sues OpenAI and Sam Altman over ChatGPT’s alleged role in violent incidents
  92. Hackers reportedly exploited Meta's AI support chatbot to take over Instagram accounts
  93. Anthropic releases Opus 4.8 and Dynamic Workflows for coordinating subagents
  94. Hark raises $700M Series A for a planned multimodal AI platform and future hardware
  95. SpaceX IPO filing details Anthropic compute deal worth about $15 billion annually
  96. Google unveils Gemini Spark, an agentic personal assistant with Gmail integration
  97. Andrej Karpathy joins Anthropic's pre-training team
  98. OpenAI launches a personal finance feature in ChatGPT with bank account connections
  99. Wirestock raises $23 million to supply multimodal training data to AI labs
  100. Google unveils Googlebook, an AI-native laptop line built around Gemini
  101. OpenAI adds a Trusted Contact safeguard in ChatGPT for possible self-harm situations
  102. OpenAI releases GPT-5.5 Instant as the new default model for ChatGPT
  103. Meta acquires Assured Robot Intelligence to support humanoid robotics AI work
  104. Pentagon signs AI deployment deals with Nvidia, Microsoft, and AWS for classified networks