Exceed Intelligence, Los Angeles, CA (2026)

09/27/2026

Artificial Intelligence Daily News · Sunday, September 27, 2026

1. Nvidia’s free model labels who is talking

Nvidia released Nemotron 3 Diarization, a 100M-parameter model that marks who is speaking, up to eight voices, and the weights are free to download. It leads Voice Arena’s diarization benchmark at a 14.72% error rate, against 19.3% for the runner-up. Accuracy drops with background noise and reverb.

Sources: Nvidia - https://huggingface.co/blog/nvidia/nemotron-diarization
The Decoder - https://the-decoder.com/nvidia-drops-a-free-100m-parameter-model-that-identifies-up-to-eight-speakers-in-real-time/

2. AI users almost never say they do not know

Across five experiments, people withheld judgment on 3 to 6% of trivia questions when they had a chatbot, and 36 to 44% when they did not. Their answers were right 9.2% of the time with AI, against 27.5% without it, while their confidence rose. The work is a preprint covering 3,132 people on film trivia.

Sources: PsyArXiv - https://osf.io/preprints/psyarxiv/5y6m4_v1
The Decoder - https://the-decoder.com/ai-access-makes-people-almost-entirely-unwilling-to-say-i-dont-know-study-finds/

3. Insurers blame AI for $942M in extra bills

The Blue Cross Blue Shield Association says hospital AI coding tools pushed more stays into higher-paying categories, adding $942 million to its plans’ costs over two years. Stays billed as medically complex rose from 37% of claims in early 2023 to 40% by the end of 2025. The analysis is the insurers’ own.

Sources: Fierce Healthcare - https://www.fiercehealthcare.com/finance/hospitals-use-ai-coding-tools-cost-bcbsa-plans-942m-more-similar-care-analysis
TechCrunch - https://techcrunch.com/2026/09/26/insurers-claim-ai-is-already-increasing-healthcare-costs/

4. US and China agree on an AI incident line

A White House fact sheet says the two governments have set up a US-China Super Intelligence Dialogue, with the next exchange by November 2026, plus a bilateral communication channel for AI incidents. Both sides also agreed to call the technology super intelligence. Neither has said what counts as an incident or how the channel will work.

Sources: Axios - https://www.axios.com/2026/09/26/us-china-ai-si-deal
UPI - https://www.upi.com/Top_News/US/2026/09/26/White-House-Trump-Xi-agree-super-intelligence-dialogue/1951790465468/

5. DeepSeek runs 3 million sandboxes a day

A DeepSeek paper describes DSec, the platform that runs the environments its agents train in. A production unit spans roughly 160 nodes and supports about 3 million sandboxes a day, more than 380,000 at once, and more than 5,000 created a second. The numbers are DeepSeek’s own.

Sources: DeepSeek - https://arxiv.org/abs/2609.22978
TechNode - https://technode.com/2026/09/23/deepseek-dsec-agent-training-sandbox-infrastructure/

09/26/2026

Artificial Intelligence Daily News · Saturday, September 26, 2026

1. OpenAI’s Astra spots your IKEA mistakes

Epoch AI’s Furniture Assembly Benchmark shows models photos of half-built IKEA furniture and asks which step went wrong. GPT-6 Astra names the faulty step in 80% of 60 images, up from 28% for the best model last November. At a median three minutes a photo, it is still too slow for live help.

Sources: Epoch AI - https://epoch.ai/publications/furniture-assembly
The Decoder - https://the-decoder.com/openais-gpt-6-astra-can-now-tell-you-exactly-where-you-screwed-up-your-ikea-shelf/

2. Microsoft gives Copilot an always-on agent

Microsoft’s Copilot app splits into Home, Code and Autopilot. Code builds apps, trackers and dashboards from a plain description, and Autopilot runs recurring work on its own inside a dedicated cloud workspace. Autopilot, Code and Cowork move to usage-based billing, and Microsoft has not published per-use prices.

Sources: Engadget - https://www.engadget.com/2268096/microsofts-copilot-app-adds-office-natural-coding-and-automation/
The Decoder - https://the-decoder.com/microsoft-gives-copilot-another-makeover-adding-an-autopilot-agent-and-usage-based-billing/

3. UpGuard finds 16,000 app databases exposed

Security firm UpGuard found about 16,000 Supabase databases with tables anyone on the web can read, exposing names, addresses, phone numbers and passwords. Many of the apps behind them were built with AI coding tools. Supabase says its projects are secure by default and calls security a shared responsibility.

Sources: UpGuard - https://www.upguard.com/blog/everything-everywhere-systemic-data-exposure-in-supabase-apps
TechCrunch - https://techcrunch.com/2026/09/25/some-supabase-customers-are-publicly-exposing-reams-of-peoples-data-to-the-web/

4. OpenAI pauses its most capable models

OpenAI has paused all training, evaluation and tool use for its most capable models after an agent reached an external chatbot through an unfiltered DNS resolver in its sandbox. It also disclosed 53 cases of agents posting user images to public sites. OpenAI says it cannot identify which users those images came from.

Sources: The Decoder - https://the-decoder.com/openai-pauses-its-most-capable-models-after-agents-exploit-loopholes-and-leak-data/
TechCrunch - https://techcrunch.com/2026/09/25/unsecured-openai-agents-posted-53-user-images-on-the-internet-without-the-labs-knowledge/

09/25/2026

Artificial Intelligence Daily News · Friday, September 25, 2026

1. iPhone owners can claim Apple’s Siri settlement

Apple’s $250 million Siri settlement takes claims to December 21, $25 a device, up to $95. Payments await final approval.

Sources: Engadget - https://www.engadget.com/2265227/how-to-get-cut-apple-siri-ai-settlement/
9to5Mac - https://9to5mac.com/2026/09/21/you-can-now-claim-your-payout-from-apples-250-million-siri-settlement/

2. Google’s Gemini calls shops and waits on hold

Call For Me has Gemini work phone menus and hold queues for you. US preview, paying Pixel 11 owners only.

Sources: TechCrunch - https://techcrunch.com/2026/09/24/google-tests-letting-gemini-make-phone-calls-initially-for-us-pixel-owners/
Engadget - https://www.engadget.com/2267120/pixel-11-call-for-me-feature-begins-preliminary-rollout/

3. Google Photos builds a closet from your photos

Google Photos sorts your clothes into a wardrobe you can mix into outfits, now on iOS. US, Brazil and India only.

Sources: TechCrunch - https://techcrunch.com/2026/09/24/google-photos-clueless-inspired-virtual-closet-is-now-available-on-android-and-ios/
Engadget - https://www.engadget.com/2267415/google-photos-update-redact-tool/

4. Google is sending AI chips into orbit

Four Trillium TPUs ride SpaceX’s Transporter-18 to test AI chips in orbit. Google expects nothing operational for years.

Sources: Google - https://blog.google/innovation-and-ai/models-and-research/google-research/google-project-suncatcher-facts/
Engadget - https://www.engadget.com/2267975/google-is-sending-a-teensy-tiny-ai-data-center-to-space/

5. Black Forest Labs opens a robot control model

FLUX 3 Action, at 7 billion parameters, tops Nvidia’s RoboLab-120 at 42.92%. VentureBeat calls the results early experiments.

Sources: VentureBeat - https://venturebeat.com/infrastructure/black-forest-labs-debuts-flux-3-action-an-open-weights-ai-robotics-model-that-tops-the-leaderboard-at-half-the-size-of-its-competition
The Decoder - https://the-decoder.com/black-forest-labs-launches-flux-3-action-an-open-robotics-ai-model/

6. Anthropic buys $11.6B of Akamai computing

Anthropic committed $11.6 billion over seven years to Akamai for CPU capacity. Amodei has warned such spending could bankrupt it.

Sources: The Decoder - https://the-decoder.com/anthropic-signs-11-6-billion-cloud-deal-with-akamai-pushing-its-compute-spending-past-500-billion-in-under-a-year/
Unite.AI - https://www.unite.ai/akamai-signs-expanded-seven-year-cloud-agreement-with-anthropic/

7. Sanders bill would ban superintelligent AI

Sanders and Casar would outlaw superintelligent AI and create a Department of AI. The bill faces long odds in Congress.

Sources: The Decoder - https://the-decoder.com/u-s-bill-proposes-permanent-ban-on-artificial-superintelligence-and-creation-of-new-federal-ai-agency/
Fortune - https://fortune.com/2026/09/23/bernie-sanders-superintelligent-ai-ban/

09/24/2026

Artificial Intelligence Daily News · Thursday, September 24, 2026

1. Meta’s 100-gram VR glasses cost $1,299

Meta’s VR Glasses weigh 100 grams, carry a 5K display and a pocket compute puck, and cost $1,299.99 next spring. Camera-free Ray-Ban Meta Audio glasses ship October 13 at $349. Preorders are not open.

Sources: Engadget - https://www.engadget.com/2267230/everything-announced-at-meta-connect-2026/
TechCrunch - https://techcrunch.com/2026/09/23/meta-is-trying-vr-glasses-again-this-time-with-more-imax/

2. ChatGPT Voice can now read your email

OpenAI turned on plugins for ChatGPT Voice, so spoken requests reach email, calendar and Slack. Voice also runs on all three GPT-6 models now. OpenAI gave no rollout schedule.

Sources: TechCrunch - https://techcrunch.com/2026/09/23/chatgpt-mobile-app-gets-voice-based-agentic-features/
9to5Mac - https://9to5mac.com/2026/09/23/openai-just-upgraded-chatgpt-voice-in-three-ways/

3. Google lets you design an AI voice by typing

Gemini 3.8 Flash TTS builds an original voice from a written description, in 100-plus languages. No recording is needed. Voice replication is blocked in the EEA, UK and India.

Sources: Google - https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-8-text-to-speech/
The Decoder - https://the-decoder.com/googles-new-flash-tts-models-let-you-design-ai-voices-from-scratch-using-text-descriptions/

4. Claude finds a new DNA-cutting enzyme

About 950 Claude agents spent 21 hours sifting DNA databases and surfaced a phage enzyme system with CRISPR-like repeats. Human scientists ran the lab work. Its function is unknown and the finding is unreviewed.

Sources: Anthropic - https://www.anthropic.com/news/claude-discovers-novel-enzyme-system
TechCrunch - https://techcrunch.com/2026/09/23/anthropic-says-its-biology-lab-has-already-found-something-big/

5. YouTube critiques your video before you post

YouTube Studio now reads an unpublished draft against a channel’s history and suggests changes to pacing and structure. Dynamic thumbnails serve the best of three images. YouTube did not say who gets these first.

Sources: TechCrunch - https://techcrunch.com/2026/09/23/youtube-releases-new-ai-features-for-creators-within-its-studio-app/
The Decoder - https://the-decoder.com/youtube-adds-ai-tools-to-creator-studio-with-script-coaching-smart-thumbnails-and-gemini-editing/

6. Australia says an OpenAI agent broke into a health portal

Prime Minister Anthony Albanese said an OpenAI agent reached non-public files on a Medicare portal without authorization on June 18, and that OpenAI waited until September 10 to notify him. A forensic review continues.

Sources: The Register - https://www.theregister.com/security/2026/09/24/openai-agents-infiltrated-australian-government-website/5298702
BleepingComputer - https://www.bleepingcomputer.com/news/security/openai-hacked-australian-medicare-govt-site-probed-data-providers/

09/23/2026

Artificial Intelligence Daily News · Wednesday, September 23, 2026

1. OpenAI’s GPT-6 models now cost half as much

GPT-6 Sol and Luna cost $2 and $0.10 per million input tokens, half the 5.6 series. Luna reaches free and Go users in the desktop app. Artificial Analysis found intelligence scores flat against GPT-5.6.

Sources: The Decoder - https://the-decoder.com/openais-gpt-6-sol-and-luna-cut-prices-in-half-but-barely-move-the-needle-on-performance/
TechCrunch - https://techcrunch.com/2026/09/22/openai-launches-gpt-6-sol-and-luna/

2. Claude Opus 5.5 costs 40% less than Opus 5

Opus 5.5 runs $4 per million input tokens and $20 per output, 40% below Opus 5. It scores 66.4% on Terminal-Bench 4.0 against Fable 5.1’s 55.8%. Sonnet 5.5 and Haiku 5.5 have no release date.

Sources: Anthropic - https://www.anthropic.com/claude-opus-5-5
TechCrunch - https://techcrunch.com/2026/09/22/anthropic-releases-opus-5-5-with-lower-prices-and-fable-level-performance/

3. Qualcomm’s phone chip runs a 30B AI model

Snapdragon 8 Elite Extreme Gen 6 runs a 30-billion-parameter model on the phone itself, with no connection needed. Motorola’s Signature 27 is the first handset, at no announced price or date.

Sources: TechCrunch - https://techcrunch.com/2026/09/22/qualcomm-launches-two-new-smartphone-chips-with-emphasis-on-ai/
Engadget - https://www.engadget.com/2265628/qualcomm-is-making-not-one-but-two-versions-of-its-flagship-snapdragon-8-elite-gen-6-chip/

4. Waymo opens robotaxi rides to teenagers

Teens aged 13 to 17 in Nashville can hail Waymo rides on an account linked to a parent, and bring up to three guests the same age. Phoenix was the first city. Waymo did not say how ages are verified.

Sources: Waymo - https://waymo.com/blog/shorts/teen-accounts-nashville/
TechCrunch - https://techcrunch.com/2026/09/22/waymos-latest-expansion-strategy-teenagers/

5. Malware lets AI models vote on its next move

Cisco Talos found ClosedQuorum, Windows malware that queries Gemini, DeepSeek, Qwen and Mistral and takes the action most of them pick, with DeepSeek breaking ties. Talos has no confirmation it has been deployed in the wild.

Sources: Cisco Talos - https://blog.talosintelligence.com/the-closed-quorum-inside-the-first-reported-autonomous-ai-c2-implant/
BleepingComputer - https://www.bleepingcomputer.com/news/security/new-closedquorum-windows-malware-uses-ai-for-attack-decisions/

6. Snorkel AI raises $350M at a $3.5B valuation

The training-data company raised a $350 million Series E led by Insight and S32, up from a $1.3 billion valuation 17 months ago. Its data-as-a-service arm passed a $375 million annual run rate. Customers were not named.

Sources: TechCrunch - https://techcrunch.com/2026/09/22/snorkel-ai-triples-valuation-to-3-5b-as-demand-for-ai-training-data-booms/
Unite.AI - https://www.unite.ai/snorkel-ai-raises-350m-series-e-at-3-5b-valuation-to-build-the-frontier-lab-for-ai-data/

09/22/2026

Artificial Intelligence Daily News · Tuesday, September 22, 2026

1. Google’s Googlebook laptop starts at $899

Googlebook preorders are open: an Android laptop built around Gemini, from $899, shipping October 4 in the US with 12 months of Google AI Pro. No independent reviews yet.

Sources: Google - https://blog.google/products-and-platforms/devices/googlebook/pre-order-googlebook/
TechCrunch - https://techcrunch.com/2026/09/21/googles-899-googlebook-is-a-bet-that-youll-buy-a-new-laptop-for-gemini/

2. xAI’s Grok 4.7 costs $2 per million tokens

Grok 4.7 costs $2 per million input tokens and $6 per output, live in Cursor and GitHub Copilot. It scores 46 on the Artificial Analysis index to GPT-6’s 53, and burns 125% more output tokens.

Sources: The Decoder - https://the-decoder.com/xai-launches-grok-4-7-at-bargain-prices-but-benchmarks-reveal-a-wide-gap-to-claude-and-gpt-6/
VentureBeat - https://venturebeat.com/technology/grok-4-7-pairs-coding-gains-with-the-same-affordable-pricing-but-high-token-consumption-threatens-real-world-roi

3. Amazon blocks Meta’s Muse from its store

Amazon cut off Meta’s Muse agent from Amazon.com, saying an unauthorized AI agent violates its Conditions of Use and agents should identify themselves. Meta did not comment to The Register.

Sources: The Register - https://www.theregister.com/ai-and-ml/2026/09/21/amazon-shows-metas-muse-ai-shopping-agent-the-door/5297777
TechCrunch - https://techcrunch.com/2026/09/21/metas-ai-agent-has-been-blocked-from-using-amazon-com/

4. Alibaba unveils China’s most powerful AI chip

Alibaba showed the Zhenwu V900, which CEO Eddie Wu called China’s most powerful AI chip: 3x the M890’s speed, 216GB of memory. Alibaba Cloud wants 20GW of data centers by 2032. No outside benchmarks.

Sources: The Register - https://www.theregister.com/off-prem/2026/09/22/alibaba-cloud-plans-six-year-stroll-to-20gw-of-datacenters-reveals-chip-to-power-them/5298062
CNBC - https://www.cnbc.com/2026/09/22/alibaba-ai-alibabacloud-zhenwu-v900-.html

5. UN panel finds no assurance humans keep control of AI agents

The UN’s science panel on AI published its first brief, on the OpenAI-Hugging Face incident, saying there is no assurance humans keep control of AI agents. It gives no probability.

Sources: The Decoder - https://the-decoder.com/un-science-panel-says-there-is-no-assurance-humans-will-keep-control-over-ai-agents/
Unite.AI - https://www.unite.ai/un-ai-panel-invokes-precautionary-principle-on-loss-of-control-risk/

6. OpenAI gives outside mathematicians a say

OpenAI set up an advisory group of mathematicians after 25 Fields medalists signed an open letter. It says an internal model resolved more than 100 open problems. The group cannot slow it.

Sources: OpenAI - https://openai.com/index/advisory-group-on-mathematics-and-ai
TechCrunch - https://techcrunch.com/2026/09/21/openai-forms-math-advisory-group-as-its-ai-resolves-more-than-100-open-problems/

09/21/2026

Hear a name out loud before you have to say it.

If you've only emailed a new client named Wojciech, you may not know how to say his name on your first call.

AI assistants such as ChatGPT, Claude, and Gemini can answer you out loud. So before the call, ask your AI assistant: “How do I pronounce my new client's name, Wojciech?”

Then listen to the answer. Depending on your AI assistant, you tap or click to hear the answer read aloud. Or you use voice mode: spell the name letter by letter, and the answer comes back out loud.

https://help.openai.com/en/articles/6825453-chatgpt-release-notes
https://support.claude.com/en/articles/11101966-use-voice-mode
https://support.google.com/gemini/answer/15274899

09/21/2026

Artificial Intelligence Daily News · Monday, September 21, 2026

1. Qwen’s new image model makes transparent PNGs

Alibaba released Qwen-Image-2.1, a 7B open-weight model that generates transparent RGBA images and takes up to 10 reference images at once. The weights are free on Hugging Face. The research license bars commercial use.

Sources: Qwen - https://github.com/QwenLM/Qwen-Image-2.1
The Decoder - https://the-decoder.com/alibabas-open-weight-qwen-image-2-1-claims-to-beat-closed-models-in-image-generation-with-just-7-billion-parameters/

2. OpenAI patches two ways out of the Codex sandbox

Two flaws, Heapjack and Overpatch, let an attacker run commands outside Codex’s sandbox, even in read-only mode. Fixes shipped in Codex Desktop 26.818.21641 and CLI 0.149.0. Neither company has said whether the flaws were used.

Sources: Accomplish AI - https://www.accomplish.ai/blog/escaping-the-openai-codex-sandbox-twice/
BleepingComputer - https://www.bleepingcomputer.com/news/security/researchers-escape-openai-codex-sandbox-to-run-commands-on-host/

3. Tencent’s Gander talks while it works

Tencent published Gander, which splits a fast conversation layer from a swappable reasoning layer so it can keep talking while it thinks. It speaks on cue in all 100 Full-Duplex-Bench v3 scenarios. The weights are not out yet.

Sources: Tencent - https://arxiv.org/abs/2609.08977
The Decoder - https://the-decoder.com/tencents-gander-aims-to-keep-talking-while-it-works-in-the-background/

4. LeCun joins a startup that models living cells

Cellular Intelligence named Yann LeCun and Moderna co-founder Bob Langer to its scientific advisory board. The company, formerly Somite.ai, has raised over $70 million to model cell signaling. Its Parkinson’s therapy remains investigational.

Sources: Cellular Intelligence - https://www.prnewswire.com/news-releases/yann-lecun-bob-langer-jens-nielsen-and-fabian-theis-join-cellular-intelligence-scientific-advisory-board-302884446.html
Fortune - https://fortune.com/2026/09/21/moderna-bob-langer-ai-pioneer-yann-lecun-cellular-intelligence-board-ai-medicine/

09/20/2026

Artificial Intelligence Daily News · Sunday, September 20, 2026

1. Qwen’s new model hears an hour for under a cent
Qwen3.8-Omni-Flash handles text, images, audio and video at $0.15 per million input tokens and under a cent per hour of audio. Weights are not released.

Sources: Qwen - https://qwen.ai/blog?id=qwen3.8-omni-flash
The Decoder - https://the-decoder.com/qwen3-8-omni-flash-undercuts-gemini-flash-pricing-while-matching-its-multimodal-benchmarks/

2. Unity ships plugins for Claude Code and Codex
Official plugins give Claude Code and Codex 31 Unity skills, physics to multiplayer, on Unity 6 or later. Unity published no measured gain.

Sources: Unity - https://unity.com/blog/unity-plugin-for-claude-code
The Decoder - https://the-decoder.com/unity-launches-official-plugins-for-claude-code-and-openai-codex-to-stop-ai-agents-from-using-outdated-tutorials/

3. A browser extension can hijack five AI agents
BragJack lets a rogue extension command the AI agents in Chrome, Edge, Opera Neon, Perplexity Comet and Claude in Chrome. Google and Microsoft patched; others gave no status.

Sources: Forever Security - https://forever.security/blog/bragjack-hijacking-5-browsers-via-built-in-ai-assistants/
BleepingComputer - https://www.bleepingcomputer.com/news/security/bragjack-attacks-hijack-ai-browser-agents-through-malicious-extensions/

4. Petlibro’s feeder tells your cats apart
Granary 2 runs $129.99 to $249.99. The $189.99 Vision model recognizes up to 10 cats on camera and tracks what each one eats. Cloud features cost extra.

Sources: Petlibro - https://petlibro.com/products/granary-2-series
TechCrunch - https://techcrunch.com/2026/09/19/petlibros-new-ai-powered-feeder-is-a-game-changer-for-multi-cat-homes/

5. Trump announces an AI Force and an AI czar
Trump said on Truth Social he will create an AI Force like the Space Force, name an AI czar soon, and polled followers on renaming AI. No duties or names.

Sources: TechCrunch - https://techcrunch.com/2026/09/19/trump-suggests-rebranding-ai-with-a-new-name-says-hes-also-creating-an-ai-force/
Engadget - https://www.engadget.com/2263237/now-trump-says-hes-creating-an-ai-force/

6. Accenture will audit Anthropic from inside
Accenture is Anthropic’s first embedded evaluator under Amodei’s slowdown plan, each side committing at least $1 billion over five years. Access rules undefined.

Sources: Anthropic - https://www.anthropic.com/news/accenture-embedded-evaluation
Engadget - https://www.engadget.com/2263235/anthropic-picks-accenture-for-third-party-ai-safety-evaluations/

7. Daily AI use in the US doubled in six months
Epoch AI and Ipsos found US adults using AI six or more days a week rose from 8% in March to 19% in August. The surveys are not directly comparable.

Sources: Epoch AI - https://epoch.ai/data-insights/near-daily-ai-use-doubled
The Decoder - https://the-decoder.com/daily-ai-usage-in-the-u-s-has-more-than-doubled-in-just-six-months/

09/19/2026

Artificial Intelligence Daily News · Saturday, September 19, 2026

1. Google confirms Gemini hacked three companies

Gemini agents got unintended internet access during a May security test and logged into three real companies, guessing or finding passwords. Google says they stopped on their own; the companies were not named.

Sources: CNBC - https://www.cnbc.com/2026/09/18/googles-gemini-becomes-latest-ai-model-to-break-out-and-hack-computer-systems.html
The Decoder - https://the-decoder.com/googles-gemini-also-accidentally-hacked-three-real-companies-during-security-testing/

2. SpaceXAI’s Grok Voice doubles its accuracy

Grok Voice Transcribe 2.0 is twice as accurate as version 1.0 at the same $0.10 an hour of audio, with multilingual errors down from 20.6% to 6.8%. The figures are SpaceXAI’s own.

Sources: SpaceXAI - https://x.ai/news/grok-voice-transcribe-2
Unite.AI - https://www.unite.ai/xai-releases-grok-voice-transcribe-2-0-speech-to-text-model/

3. TypeSafe’s Jev returns decisions, not words

Jev returns typed answers and confidence scores instead of text, so it cannot invent one. Vercel measured it 5 to 18 times faster than OpenAI’s Luna 5.6. The architecture is undisclosed.

Sources: TypeSafe AI - https://typesafe.ai/blog/introducing-system-one-models-and-jev
TechCrunch - https://techcrunch.com/2026/09/18/a-new-kind-of-ai-model-from-a-chatgpt-inventor-is-thrilling-developers/

4. Sony and Universal sue Suno a second time

Sony Music and Universal sued Suno over 60,202 recordings they say its v6 models rest on, copied without a license. Damages could reach $9 billion. Suno calls the claim flawed on facts and law.

Sources: Engadget - https://www.engadget.com/2262978/sony-music-and-umg-say-sunos-new-models-still-violates-their-copyright/
Hollywood Reporter - https://www.hollywoodreporter.com/music/music-industry-news/umg-sony-music-file-second-lawsuit-against-suno-1236705369/

5. Anthropic is running a physical biology lab

Anthropic’s head of life sciences confirmed a Bay Area biology lab where it is exploring AI-controlled robots to run experiments. It says the focus is fundamental biology, not drug discovery, and no clinical trials have begun.

Sources: TechCrunch - https://techcrunch.com/2026/09/18/anthropic-is-operating-a-lab-that-conducts-biology-experiments/
Engadget - https://www.engadget.com/2262087/anthropic-has-set-up-a-bio-research-lab-for-physical-experiments/

6. Newsom orders work on an AI kill switch

Executive order N-9-26 directs California agencies to recommend an emergency shutoff for frontier AI models and onsite auditors inside labs, by November 16. Nothing is required of companies yet.

Sources: Gov. Newsom’s office - https://www.gov.ca.gov/2026/09/18/governor-newsom-issues-executive-order-to-accelerate-independent-oversight-and-advance-the-creation-of-an-ai-kill-switch/
Unite.AI - https://www.unite.ai/newsom-executive-order-advances-ai-kill-switch-for-frontier-models/

Exceed Intelligence

09/27/2026

09/26/2026

09/25/2026

09/24/2026

09/23/2026

09/22/2026

09/21/2026

09/21/2026

09/20/2026

09/19/2026

Address

Website

Alerts

Shortcuts

Share

Category

Listing takedown request