# AI news brief: Friday, November 13, 2026

37 stories, most important first. Status is how solid each one is: official, confirmed, reported or rumor.

- [Nvidia CEO Jensen Huang resigns; CFO Colette Kress becomes interim CEO](https://ai-news.ikangai.com/story/nvidia-ceo-jensen-huang-resigns-cfo-colette-kress-becomes-interim-ceo/index.md): reported. NVIDIA says Jensen Huang has stepped down as CEO, president and board member after 33 years, with CFO Colette Kress serving as interim CEO.
- [OpenAI begins embedded assessments with Apollo Research, METR, Transluce and Goodfire](https://ai-news.ikangai.com/story/openai-begins-embedded-assessments-with-apollo-research-metr-transluce-and-goodf/index.md): official. OpenAI says it has started the first phase of embedded assessments by outside evaluators of its model alignment, safeguards and monitoring.
- [OpenAI launches ChatGPT Wallet so agents can make purchases with user approval](https://ai-news.ikangai.com/story/openai-launches-chatgpt-wallet-so-agents-can-make-purchases-with-user-approval/index.md): reported. OpenAI says ChatGPT Wallet lets agents in ChatGPT Work, Codex and dots complete purchases once the user approves them.
- [Anthropic opens private beta for self-hosted Claude Managed Agents](https://ai-news.ikangai.com/story/anthropic-opens-private-beta-for-self-hosted-claude-managed-agents/index.md): reported. Anthropic's self-hosted Claude Managed Agents private beta lets customers run the agent harness, sandboxes and session history in their own cloud or on-prem.
- [DHS and FBI examine whether an FAA AI tool or software update caused a ground stop](https://ai-news.ikangai.com/story/dhs-and-fbi-examine-whether-an-faa-ai-tool-or-software-update-caused-a-ground-st/index.md): reported. DHS says it is examining the FAA's SMART system after a nationwide ground stop, and the FBI is reportedly looking at a software update.
- [FT: Nvidia's board approached Arm CEO Rene Haas about replacing Jensen Huang](https://ai-news.ikangai.com/story/ft-nvidia-s-board-approached-arm-ceo-rene-haas-about-replacing-jensen-huang/index.md): reported. The Financial Times reports that Nvidia's board approached Arm CEO Rene Haas as one of several candidates to succeed Jensen Huang, and that no offer has been made.
- [Fed Governor Michael Barr warns AI could cause short-term labor disruption](https://ai-news.ikangai.com/story/fed-governor-michael-barr-warns-ai-could-cause-short-term-labor-disruption/index.md): reported. Federal Reserve Governor Michael Barr says society should prepare for serious short-term labor disruption as AI spreads through the economy.
- [SpaceX plans billions for supercomputing campuses in Texas and Arizona](https://ai-news.ikangai.com/story/spacex-plans-billions-for-supercomputing-campuses-in-texas-and-arizona/index.md): reported. SpaceX says it plans to invest billions in supercomputing campuses at Starbase, Texas, and in Arizona.
- [CoreWeave, Nebius fund AI buildouts with customer prepayments and contract-backed loans](https://ai-news.ikangai.com/story/coreweave-nebius-fund-ai-buildouts-with-customer-prepayments-and-contract-backed/index.md): reported. CoreWeave and Nebius are financing data center expansion with customer prepayments and contract-backed loans, while Nebius shares fell 17% after a delivery delay.
- [Yahoo becomes first licensee of Internet Licensing Collective's pay-to-scrape model](https://ai-news.ikangai.com/story/yahoo-becomes-first-licensee-of-internet-licensing-collective-s-pay-to-scrape-mo/index.md): reported. Yahoo is the first licensee of the Internet Licensing Collective, which requires AI companies to pay for content access through collective licenses.
- [Reported rule would restrict Chinese access to overseas Blackwell-class AI data centers](https://ai-news.ikangai.com/story/reported-rule-would-restrict-chinese-access-to-overseas-blackwell-class-ai-data/index.md): reported. A reported rule would limit Chinese access to overseas AI data centers for future exports of Blackwell-class or better chips in some countries.
- [Baidu says ERNIE 5.5 Flash Preview has launched, with weights due November 25](https://ai-news.ikangai.com/story/baidu-says-ernie-5-5-flash-preview-has-launched-with-weights-due-november-25/index.md): reported. Baidu says its ERNIE 5.5 Flash Preview has launched, and model weights are expected by November 25.
- [Claude Code 2.1.325 patches Bash permission gaps and a zstd decoding flaw](https://ai-news.ikangai.com/story/claude-code-2-1-325-patches-bash-permission-gaps-and-a-zstd-decoding-flaw/index.md): official. The changelog lists fixes for compound Bash commands that could run without asking, an allow rule that could run a different program, and a zstd decoding security issue.
- [Theo says Gemini 4 Carbon runs about five times slower than other models in his work](https://ai-news.ikangai.com/story/theo-says-gemini-4-carbon-runs-about-five-times-slower-than-other-models-in-his/index.md): reported. Commentator Theo says his own audit found Gemini 4 Carbon about five times slower than comparable models, with over 11 seconds before responses start.
- [Anthropic's Amodei tells investors AI could end major diseases within a decade](https://ai-news.ikangai.com/story/anthropic-s-amodei-tells-investors-ai-could-end-major-diseases-within-a-decade/index.md): reported. Dario Amodei told prospective investors AI could make cancer, Alzheimer's, heart disease and mental illness relics within a decade.
- [Survey: most firms that replaced a production AI model spent weeks on it](https://ai-news.ikangai.com/story/survey-most-firms-that-replaced-a-production-ai-model-spent-weeks-on-it/index.md): reported. A survey finds 69% of firms that replaced a production AI model spent weeks on the switch, and 36% needed over a month.
- [Chinese regulators penalize firms that flood the web with fake product rankings](https://ai-news.ikangai.com/story/chinese-regulators-penalize-firms-that-flood-the-web-with-fake-product-rankings/index.md): reported. Chinese regulators are penalizing firms that flood the web with fake rankings meant to sway chatbot shopping recommendations.
- [Consulting firms test outcome-based pricing as AI erodes billable hours](https://ai-news.ikangai.com/story/consulting-firms-test-outcome-based-pricing-as-ai-erodes-billable-hours/index.md): reported. Consulting firms are testing outcome-based pricing and reshaping graduate roles as AI automates junior work.
- [48 Google AI agents flag possible unmapped Maya site in Mexico from NASA LiDAR](https://ai-news.ikangai.com/story/48-google-ai-agents-flag-possible-unmapped-maya-site-in-mexico-from-nasa-lidar/index.md): reported. Forty-eight Google AI agents scanned NASA LiDAR data and flagged a possible unmapped Maya site in Mexico, which is now under investigation.
- [United Airlines blocks AI agents from booking flights over security concerns](https://ai-news.ikangai.com/story/united-airlines-blocks-ai-agents-from-booking-flights-over-security-concerns/index.md): reported. United Airlines is blocking AI agents from booking flights, citing risks from shared credentials and federal rules that require human check-in confirmations.
- [Connecticut requires employers to disclose AI use in hiring from October 2027](https://ai-news.ikangai.com/story/connecticut-requires-employers-to-disclose-ai-use-in-hiring-from-october-2027/index.md): reported. Connecticut employers that use AI in hiring or workforce decisions must disclose that use to employees and applicants starting October 2027.
- [OpenAI launches Daybreak Defense Factory and Daybreak Red Team cyber tools](https://ai-news.ikangai.com/story/openai-launches-daybreak-defense-factory-and-daybreak-red-team-cyber-tools/index.md): reported. OpenAI has launched Daybreak Defense Factory and Daybreak Red Team, two cyber products, according to a social media post.
- [Perplexity releases Decider v1.3 open-weight models with a span-locating task](https://ai-news.ikangai.com/story/perplexity-releases-decider-v1-3-open-weight-models-with-a-span-locating-task/index.md): reported. Perplexity's Decider v1.3 adds a locate task that returns span offsets, and the company says it beats its production classifiers by 6.54 points.
- [Salesforce's SOV-Transcribe-1 tops open-weights speech-to-text at 2.3% AA-WER](https://ai-news.ikangai.com/story/salesforce-s-sov-transcribe-1-tops-open-weights-speech-to-text-at-2-3-aa-wer/index.md): reported. Artificial Analysis ranks Salesforce's SOV-Transcribe-1 as the top open-weights speech-to-text model at 2.3% AA-WER with 4.6B active parameters.
- [Codex CLI 0.169.0 adds multiple side-chat tabs to its opt-in terminal workspace](https://ai-news.ikangai.com/story/codex-cli-0-169-0-adds-multiple-side-chat-tabs-to-its-opt-in-terminal-workspace/index.md): official. OpenAI's Codex CLI 0.169.0 lets users run several side chats as tabs in the opt-in terminal workspace and reopen saved local side chats after a restart.
- [Codex exec keeps project config and MCP servers off in untrusted directories](https://ai-news.ikangai.com/story/codex-exec-keeps-project-config-and-mcp-servers-off-in-untrusted-directories/index.md): official. Codex CLI 0.169.0 requires explicit project trust for codex exec, so project configuration and MCP servers stay disabled in untrusted directories.
- [Codex CLI 0.169.0 lets users add custom model providers and sign in with a device code](https://ai-news.ikangai.com/story/codex-cli-0-169-0-lets-users-add-custom-model-providers-and-sign-in-with-a-devic/index.md): official. Codex CLI 0.169.0 adds custom model provider editing in the terminal and a device-code sign-in option for supported gateways over SSH or without a local browser.
- [Codex CLI 0.168.0 lets users pick a default model provider for new chats](https://ai-news.ikangai.com/story/codex-cli-0-168-0-lets-users-pick-a-default-model-provider-for-new-chats/index.md): official. OpenAI's Codex CLI 0.168.0 adds a model provider browser to the TUI and a setting for the default provider used by new chats.
- [Codex CLI 0.168.0 adds opt-in side conversation panes to the terminal UI](https://ai-news.ikangai.com/story/codex-cli-0-168-0-adds-opt-in-side-conversation-panes-to-the-terminal-ui/index.md): official. Codex CLI 0.168.0 adds an opt-in workspace mode that shows a side conversation beside the main chat, with resizable and zoomable panes.
- [Codex CLI 0.168.0 adds opt-in animated GIF generation for image tasks](https://ai-news.ikangai.com/story/codex-cli-0-168-0-adds-opt-in-animated-gif-generation-for-image-tasks/index.md): official. Codex CLI 0.168.0 lets users enable features.image_generation_animation to generate or edit animated GIFs with frame count and timing controls.
- [Codex CLI 0.168.0 deprecates the multi_agent flag in favour of agents.enabled](https://ai-news.ikangai.com/story/codex-cli-0-168-0-deprecates-the-multi-agent-flag-in-favour-of-agents-enabled/index.md): official. Codex CLI 0.168.0 deprecates the features.multi_agent flag and warns users who set it explicitly to use agents.enabled to turn off subagent tools.
- [Claude Code 2.1.325 stops sandboxed Bash from running without prompts by default](https://ai-news.ikangai.com/story/claude-code-2-1-325-stops-sandboxed-bash-from-running-without-prompts-by-default/index.md): official. Anthropic's Claude Code 2.1.325 makes sandboxed Bash commands follow the normal permission flow unless sandbox.autoAllowBashIfSandboxed is set to true.
- [Claude Code 2.1.325 makes scheduled tasks ask before running some skills](https://ai-news.ikangai.com/story/claude-code-2-1-325-makes-scheduled-tasks-ask-before-running-some-skills/index.md): official. Scheduled tasks and /loop now ask before running a skill that declares allowed-tools or hooks, and Skill(name) must be added to permissions.allow to run it unattended.
- [Claude Code 2.1.325 asks before read-only git commands in untrusted repositories](https://ai-news.ikangai.com/story/claude-code-2-1-325-asks-before-read-only-git-commands-in-untrusted-repositories/index.md): official. In repositories that are not the project's or that the user has not trusted, read-only git commands now ask first, and --add-dir is used to allow them without prompts.
- [Claude Code 2.1.325 applies context window setting to unrecognised claude- model IDs](https://ai-news.ikangai.com/story/claude-code-2-1-325-applies-context-window-setting-to-unrecognised-claude-model/index.md): official. CLAUDE_CODE_MAX_CONTEXT_TOKENS now also sets the context window for unrecognised model IDs that start with claude-, such as a gateway's claude-sonnet-latest.
- [Claude Code 2.1.325 refuses marketplace pins to commits that older clients cannot clone](https://ai-news.ikangai.com/story/claude-code-2-1-325-refuses-marketplace-pins-to-commits-that-older-clients-canno/index.md): official. Marketplace ref now accepts the full ID of any commit on a branch or tag, and clients before 2.1.325 cannot clone such a pin, so those marketplaces stop loading there.
- [Claude Code 2.1.325 skips plugins' unsafe paths and links and warns about plugin code](https://ai-news.ikangai.com/story/claude-code-2-1-325-skips-plugins-unsafe-paths-and-links-and-warns-about-plugin/index.md): official. Plugin loading now skips some unsafe paths and links in a plugin's folder, claude plugin validate reports them, and plugin prompts say plugins can run code on the computer.

---

AI wrote this summary from the sources listed, and nobody reviewed it individually before publication. Check the original source before you rely on it.
