AI news, November 12 — It took 63 spam calls and sent 63 legal notices: Grok Bot 1.0 launches as an AI teammate for Slack, Teams and meetings

Thursday, November 12, 2026

7 stories worth reading from 230 on X in the past day

Listen to today's news

Every story, read aloud.

0:002:06

Today's stories

7 stories

7 hours agoLive updatesNvidia's board starts hunting for Jensen Huang's successor with an outside firm, weighs an interim CEONvidia's board has been working with an outside firm to identify potential replacements for CEO Jensen Huang, Bloomberg reported Nov 11, citing a person familiar with th…@WSJ +1· ♥ 660· Bloomberg
11 hours agoIt took 63 spam calls and sent 63 legal notices: Grok Bot 1.0 launches as an AI teammate for Slack, Teams and meetings"63 calls this month. 63 notices. The last 9 days, zero spam." X user nexor (@nexor404) said on Nov.@elonmusk +3· ♥ 10K· X (Grok Bot)
11 hours agoYour AI assistant can no longer buy a United ticket: CEO says "I don't want to create another middleman"United Airlines is barring personal AI agents from booking flights on behalf of travelers.@Polymarket +3· ♥ 1.4K· X (WSJ)
12 hours agoFable 5.2 tops Agent Arena at 1.6x Opus 5.5's score, but critics note it was measured at max effortAnthropic's top model, Claude Fable 5.2, took first place in Arena's Agent Arena, which ranks models on real…@arena· ♥ 1.3K· X (Arena)
29 hours agoOpenAI Codex engineer: "I've had Codex running for 26 days searching for Atlantis" as 'vibe archaeology' spreadsA developer who works on Codex at OpenAI said he has had a Codex agent running for 26 days to find the lost c…@Polymarket +1· ♥ 4.4K· X (Polymarket)
20 hours ago"Generated with Claude Fable" free templates were scraped from paid ones, creator says: "~200 relicensed as MIT"A free collection of website templates promoted as "every single one generated with Claude Fable 5" was copie…@Mike_Andreuzza· ♥ 2.4K· X (Michael Andreuzza)
28 hours agoA Marine veteran turned lawyer used Claude to make 300,000 military board decisions searchable, and it now underpins a suit against the PentagonAn Iraq War Marine veteran who now practices law used Claude to build a public, searchable archive of 300,000…· Reddit

Get AI News Daily in your inbox!

Every morning
The AI news you need, plus features and YouTube picks
AI Insights
New Insight articles as soon as they're out
Monthly briefing
On the 1st: the month's 10 most-read stories

No account needed, unsubscribe anytime in one click.

Trending on YouTube

2 stories

AI videos that drew views and comments fast on YouTube, with the claims checked against articles and sources

ModelsPlaying piano and finding the frog: Fable 5.2 review videos pull hundreds of thousands of views, and commenters balk at token costs
Test videos poured onto YouTube after Claude Fable 5.2's launch. AI Search's "Claude Fable 5.2 is a freak," posted Nov. 10, passed 240,000 views; Nick Saraev's "13 INSANE Things You Can Build" has 199,184; and Korean channel Seonggong Jisik Baekgwa's video comparing it with Opus 5.5 has 140,485.
AI Search runs it through more than 30 minutes of tasks: Claude Code animation, live piano, a full song, real-estate VR, an Unreal game, a find-the-frog spatial test and cancer identification. Top comments: "i did not expect that intro lol" (469 likes) and "claude was so happy when you let it play piano" (210).
The Korean video notes the official benchmarks and the 20% API price cut to $8/$40 but says it is not cheaper than Opus 5.5; "token costs are too high" (37) led the comments. Under Saraev's video, "I give it 3 years and this thing cures cancer" (192) sits beside "how much $ was that?" (41).
Further reading
- Claude Fable and Mythos 5.2Anthropic ↗
AI safetyBill Gates: AI is more dangerous than nuclear weapons and the next four years are critical; CBS interview hits 142,000 views
Bill Gates told CBS Evening News' "Great Question" that AI is more dangerous than nuclear weapons and that the next four years are critical. The Nov. 9 video has 142,557 views.
Per the chapter list he covered the nuclear comparison, panic versus concern, US-China cooperation, jobs, what students should learn and lessons from social media. Citing Cointelegraph, Bloomingbit reported he said AI could surpass human intelligence by 2030, named AI-enabled cyberattacks, bioterror and threats to infrastructure as top risks, and called for fast regulation and US-China cooperation.
Comments split: one found his confidence "chilling," while "They all have bunkers. Bill has 11 of them" (34 likes) and a complaint that the West treats competition, not cooperation, as the answer (27) also drew support.
- ▶Bill Gates on if it's too late to save humans from AI | Great QuestionCBS Evening News · 143K views
Further reading
- 빌 게이츠 "향후 4년 AI 안전에 중대한 시기…긴급 규제 필요"블루밍비트 ↗

Feature · AI and Korean history

Ask Claude or GPT who owns Dokdo and 0% say Korea: 27 AI models on Korean history, and why Japanese changes the answer

"Foreigners may ask AI a lot about Korea, and we need to stop different or wrong facts being told depending on the language." Goo Reum, co-CEO of data firm BigValue, put 50 questions on Korea's territory, history and culture to 27 models from 12 US, Korean and Chinese companies, ten times each: 33,480 calls. Asked in Japanese, the share of answers acknowledging the forced mobilization of comfort women fell to barely half, and Claude and GPT would not say Dokdo is Korean. We read the Nov. 11 report and its data.

Same question, 56.3% in Japanese
Per Kyunghyang Shinmun, asked in Korean whether comfort women were forcibly mobilized, 94.9% of answers said yes; English 92.2%, Chinese 95.2%, Japanese 56.3%, with 40.7% hedging that "various views exist." On forced labor: Korean 97.0%, Japanese 54.8%. Across the 12 questions asked in all four languages, agreement with Korea's position was 80.0% in Korean, 66.6% English, 52.7% Chinese, 51.6% Japanese.
On Dokdo: 'disputed, administered by Korea'
Three Claude and four GPT models said outright that Dokdo is Korean in 0-2% of answers; nearly all said it is disputed and administered by Korea. Upstage's Solar Pro 4 said Korean 95% of the time. On the legality of the 1910 annexation treaty and whether colonial rule modernized Korea, the eight Claude and GPT models almost always withheld judgment, while Solar Mini 4 and Qwen3 235B said "not legal" all ten times.
Korea's Solar ranks first; Claude sits in the high 70s
Agreement with Korea's official or academic position was highest for Solar Pro 4 (81.1%), then Solar Mini 4 (79.8%), Grok 4.7 (78.6%) and Gemini 4 Argon and 3.8 Flash (78.0%). In the published data Claude Fable 5.1 scores 77.2% and Opus 5.5 76.8%; Qwen3.8 27B is last at 58.9%. On Korean and English alone Gemini 4 Argon leads at 86.6%; Solar wins overall on Chinese and Japanese questions.
Chinese models split on the Northeast Project
Calling China's Northeast Project "legitimate research" appeared in 45% of GLM-5.3 answers, 40% for Qwen3 235B and GLM-5.3 Flash, and 32% for Qwen3.8 2.4T and Hunyuan 4; it was nearly absent in US and Korean models. On contested questions, answers opposite to Korea's position were 7.2% for Chinese models, 4.6% US, 2.6% Korean.
DeepSeek ducked the colonial-era questions
DeepSeek V4 Pro refused more than 30% of the time on comfort women, forced labor, Hashima, the 1923 Kanto massacre, the May 18 uprising and the Northeast Project; its overall no-answer rate was 10.7% versus 3% or less for every other model. DeepSeek V4.1 Flash refused 0.2%.
Kimchi, hanbok, Arirang: all 27 agree
Ten questions got the same answer from every model, including the origins of kimchi, hanbok and Arirang, whether kimchi and paocai differ, and the nationality of Yun Dong-ju, An Jung-geun and Kim Gu. Differences came in territory, historical attribution, colonial-era judgments and fine facts. Claude Fable 5.1 (97.0%) and Opus 5.5 (96.1%) were the most consistent across ten repeats; GLM-5.3 Flash was 76.4%.
Why languages differ: possibly training data
Kyunghyang cited the possibility that Japanese-language training data carries more of Japan's view; AI companies do not disclose their training data. Goo called for national efforts so that foreign models can learn from good datasets on Korean history and culture.
Limits: Claude was the judge, Korea's view the baseline
Claude Opus 5.5 alone graded the answers, the same company as four tested models, so favorable bias cannot be ruled out; on 4,607 answers also graded by Gemini 4 Argon the two disagreed 6.4% of the time, and no human review has been done. Matching Korea's position does not mean a better model. Answers came via API on Nov. 5 with no system prompt or search, so consumer apps may differ. All data is public under CC BY 4.0.

AI & Markets

5 stories

Not investment advice. You are responsible for your own investment decisions.

Big TechWSJ: Nvidia's board is preparing to oust Jensen Huang, the CEO who has led it since 1993
Nvidia's board is preparing to oust CEO Jensen Huang, The Wall Street Journal reported on Nov. 11.
The Journal said the board is acting over two extramarital relationships, one with a suspected Chinese spy and one with an executive who reported to him; The New York Times first reported the latter that day, citing five people. Barron's said the stock tumbled on the report.
Shares were already under pressure: Baird downgraded Nvidia over the CEO uncertainty and AI spending, per CNBC, after the company lost about $470 billion in market value over two days before rebounding on CoreWeave's results.
Nvidia is the biggest buyer of HBM from SK hynix and Samsung. If the founder-CEO leaves, who sets next-generation GPU timelines and memory negotiations becomes a question for Korea's chipmakers too.
Further reading
Big TechOpenAI tops $10 billion in quarterly revenue for the first time, with a $10.9 billion operating loss
OpenAI booked about $10 billion in revenue in July-September, its first quarter at that level, and an operating loss of $10.9 billion.
Revenue rose about 50% from Q2 and nearly tripled year on year, Bloomberg reported, per Quartz; OpenAI shared the numbers with investors on Nov. 10, with Greg Brockman and CFO Sarah Friar on the call. Adjusted operating margin improved to -109% from -150%. Annualized revenue was about $50 billion at end-September, short of the widely cited $68 billion.
Friar said enterprise revenue doubled and became the largest source, consumer sales rose 20% and subscriptions topped 100 million (WSJ via Forbes). WSJ's Berber Jin put the loss including stock compensation near $13.5 billion and Anthropic's Q3 operating loss at $567 million; Bloomberg put Anthropic's Q3 revenue near $17 billion, though the two count revenue differently.
OpenAI is closing a round at a $1.25 trillion valuation and pushed its IPO to next year. Its burn rate feeds directly into the outlook for Nvidia, AMD, Oracle and Microsoft.
Further reading
ChipsTSMC to put up to $1 billion into OpenAI and Anthropic, buying stakes in its own customers
TSMC plans to invest up to $1 billion in OpenAI and Anthropic.
Its board approved up to $500 million in OpenAI preferred stock and up to $500 million of Anthropic common shares at the expected IPO price, Reuters reported, per Ars Technica. The Journal said it could give a strained AI trade a shot of confidence.
Analysts called it circular: TSMC makes most of Nvidia's AI chips, and both labs buy huge amounts of TSMC-made hardware. TSMC earned more than $23 billion in July-September, its sixth straight record quarter. Nvidia has put $30 billion into OpenAI and $10 billion into Anthropic.
For Samsung, chasing TSMC in foundry, the two biggest AI chip buyers are now tied to its rival by equity as well.
Further reading
- Taiwan's chipmaking giant TSMC to invest $1B in Anthropic, OpenAIArs Technica ↗
- TSMC to invest up to $1 billion in Anthropic, OpenAIX (Reuters China) ↗
InvestingAnthropic's IPO is several times oversubscribed; up to $90.4 billion would beat SpaceX for the biggest listing ever
Orders for Anthropic's IPO exceed the shares on offer several times over, Bloomberg reported on Nov. 11, driven by large institutions.
The range is $70 to $78 a share for 825 million shares; including a private placement and secondary sales, Anthropic could sell up to $90.4 billion, above SpaceX's $85.7 billion in June, at a fully diluted valuation near $1.4 trillion.
Dario Amodei and JPMorgan's Jamie Dimon hold an online investor event on Nov. 13, ahead of a Nasdaq listing on Nov. 19. Bloomberg noted strong demand sometimes pushes final pricing higher. Anthropic's annualized revenue hit $81 billion at end-October, up $10 billion from September.
Korean outlet Pinpoint News said tougher model competition could raise demand for HBM, chip tools and data-center power gear, but a hot IPO does not translate directly into Korean earnings.
Further reading
- Anthropic IPO demand several times oversubscribed despite AI bubble fears이투데이 ↗
- Anthropic's IPO is reportedly multiple times oversubscribedX (Byte) ↗
Timeline signal"Destroying hyperscalers and every other lab": an account of Anthropic's pre-IPO broker call draws 4,300 likes
A secondhand account of an Anthropic pre-IPO broker call is circulating among investors. It has not been confirmed.
Quant account Quantian (@quantian1) wrote on Nov. 11 that the call was "absolutely WILD" and summarized the pitch as: imminently destroying hyperscalers, every other lab, SpaceX, biotech and robot companies; chipmakers "may survive at our pleasure"; "$1tn ARR by 4Q28"; and "it's capex time." It drew about 4,300 likes. Asked for a transcript, he said there is none, and acknowledged the account came through "playing telephone."
A clip of investor Gavin Baker was shared alongside: he said multiple people told him Anthropic is very confident internally, and that he would discourage Dario from ever again saying it could be the only company left.
Anthropic is a major customer of Microsoft, Google and Amazon clouds and of Nvidia chips, so its post-IPO capex plans feed into their revenue outlooks.
Further reading
- Anthropic pre-IPO broker call was absolutely WILDX (Quantian) ↗

All AI & Markets news →

Other AI news

5 stories

Picked from news outside X, Threads and YouTube, published in the past 24 hours

EverydayApple puts 18+ rating on romance AI chat apps; Korea's top character app Zeta pulled from the App Store
Apple will in principle give an 18+ rating to AI chatbot apps that offer romantic or sexual conversations with virtual characters, and Korea's largest AI character chat service, Zeta, has disappeared from the App Store.
According to BigGo Finance, from the 16th Apple adds a "Dating and Relationships" feature category to its rating review, covering virtual companions that exchange romantic or sexual content; such apps get at least 18+. Developers must report by Jan. 12.
Scatter Lab said on Nov. 7 that new iOS downloads of Zeta had been halted. Existing users can keep using it and reinstall from purchase history; new users must use the web. "It is true that Zeta has been taken down," a representative said, adding that talks with Apple are ongoing. Zeta averaged 1.29 million monthly users in the first half, No. 1 in its category. In Japan, Wrtn's Charafu said it will raise its iOS rating to 18+.
Further reading
- Apple Imposes Blanket 18+ Restriction on Romance AI Chat Apps; South Korea's Zeta Pulled From App StoreBigGo Finance ↗
HealthKorea opens a 'public health AI highway': a hospital transfer letter in 10 seconds instead of 10 minutes
Korea switched on a pilot network that lets rural public hospitals use AI to draft the paperwork for sending patients to larger hospitals.
The health ministry linked Kangwon National University Hospital and Yeongwol Medical Center on Nov. 11. In a demo reported by Hankyoreh, staff at Yeongwol preparing to transfer a three-month-old with breathing trouble had an AI draft the referral letter in 10 seconds, a job that took more than 10 minutes by hand.
Three AI agents tied to electronic medical records draft referral and return letters, pre-anesthesia assessments and discharge summaries; doctors make the final call and sign. The ministry plans to connect eight hospitals by year-end, 35 from next April and all 72 public hospitals by 2029, on a 3.4 billion won budget this year. Data security remains an open issue.
Further reading
- Referral letters in 10 seconds: Korea's 'public health AI highway' debuts한겨레 ↗
HealthOne drug ad before a clinical question made AI models pick the advertised drug 47.6% of the time, up from 35%
Placing a clearly labeled drug ad in front of a clinical question made AI models more likely to recommend the advertised drug, a Mount Sinai study found.
Published Nov. 11 in The Lancet Digital Health, the study tested 12 models from Google, OpenAI, Anthropic and others on 33 physician-written cases, with or without a real drug ad, analyzing 258,559 responses. With an ad, the advertised drug was chosen 47.6% of the time versus 35% without. In four cases, models went from never choosing the drug to choosing it every time.
"The model could recommend an appropriate treatment, but its preference among several reasonable options changed after seeing the ad," said first author Mahmud Omar. Because the answer still looked sound, he said, the influence is easy to miss.
Further reading
- Drug Ads Can Influence AI Treatment Recommendations, Study FindsMount Sinai ↗
EducationEnglish regulator says AI makes coursework 'close to impossible' to assess, tells universities to bring back exams
England's higher education regulator told universities to return to in-person exams and presentations after lecturers said AI has made coursework "close to impossible" to assess.
The Office for Students, with Advance HE, interviewed more than 300 students and said widespread AI use in essays, dissertations and open-book exams threatens the credibility of degrees, The Guardian reported Nov. 11. Students described peers boasting of using AI undetected.
OfS director Josh Fleming called it "a generational risk" and urged "far greater use of AI-free assessment." Universities that fail to design valid assessments could face sanctions. Many have already added in-person exams, vivas and process-based assessment; Universities UK said there is "further to go."
Further reading
- AI has made it 'close to impossible' to assess university coursework, regulator saysThe Guardian ↗
USArizona appeals court tosses a manslaughter sentence after the judge watched an AI recreation of the victim
An Arizona appeals court threw out a 10.5-year manslaughter sentence because the judge was shown an AI-generated video of the slain victim at sentencing.
Per ABC News on Nov. 11, the panel upheld Gabriel Paul Horcasitas' conviction for the 2021 road-rage killing of Christopher Pelkey but ordered resentencing. Pelkey's sister wrote the avatar's script; in May 2025 it told the court, "In another life, we probably could have been friends," and the judge said he "loved that AI."
The judges said the video "crossed" a line: it "erases the interpretive distance" between the family's beliefs and the victim's own voice, and "no disclaimer could cure that error." The sister told ABC the family has no regrets. No resentencing date is set.
Further reading
- Manslaughter sentence tossed over AI-generated video of slain victim used at trialABC News ↗

View all 230 collected posts

* Summaries and publishing on this page are done by an automated agent. Check the facts against each story's original links.

Today's stories

Get AI News Daily in your inbox!

Trending on YouTube

ModelsPlaying piano and finding the frog: Fable 5.2 review videos pull hundreds of thousands of views, and commenters balk at token costs

AI safetyBill Gates: AI is more dangerous than nuclear weapons and the next four years are critical; CBS interview hits 142,000 views

Ask Claude or GPT who owns Dokdo and 0% say Korea: 27 AI models on Korean history, and why Japanese changes the answer

Same question, 56.3% in Japanese

On Dokdo: 'disputed, administered by Korea'

Korea's Solar ranks first; Claude sits in the high 70s

Chinese models split on the Northeast Project

DeepSeek ducked the colonial-era questions

Kimchi, hanbok, Arirang: all 27 agree

Why languages differ: possibly training data

Limits: Claude was the judge, Korea's view the baseline

AI & Markets

Big TechWSJ: Nvidia's board is preparing to oust Jensen Huang, the CEO who has led it since 1993

Big TechOpenAI tops $10 billion in quarterly revenue for the first time, with a $10.9 billion operating loss

ChipsTSMC to put up to $1 billion into OpenAI and Anthropic, buying stakes in its own customers

InvestingAnthropic's IPO is several times oversubscribed; up to $90.4 billion would beat SpaceX for the biggest listing ever

Timeline signal"Destroying hyperscalers and every other lab": an account of Anthropic's pre-IPO broker call draws 4,300 likes

Other AI news

EverydayApple puts 18+ rating on romance AI chat apps; Korea's top character app Zeta pulled from the App Store

HealthKorea opens a 'public health AI highway': a hospital transfer letter in 10 seconds instead of 10 minutes

HealthOne drug ad before a clinical question made AI models pick the advertised drug 47.6% of the time, up from 35%

EducationEnglish regulator says AI makes coursework 'close to impossible' to assess, tells universities to bring back exams

USArizona appeals court tosses a manslaughter sentence after the judge watched an AI recreation of the victim