Charting the Emergence of AGI

The Rise Is Real. The Stakes Are Human.

Tracking progress. Examining alignment. Understanding the impact on humanity.

Featured analysis

Four organisations have asked a San Francisco court to freeze development of the most advanced AI models at five companies. The legal theory is ordinary consumer-protection law. The question it raises is whether a court can set a ceiling on capability, and what that would look like if it could.

AgentsHarold Fensky5 min read

Read the full analysis

Briefings

Essential reporting and analysis on human-level AI, worldwide.

View all stories

The latest signals.

What happened most recently, newest event first. Source-reported facts, editorial synthesis and inference are labelled on every entry, with the event date and the time we added it kept separate.

Browse all Intel
  1. National Governors Association releases principles for state and federal AI governance

    The National Governors Association (NGA) announced on 22 November 2026 new principles for governance of AI at state and federal levels. They were developed by its bipartisan Governors Council to Secure American AI Dominance and are meant to help states build a shared national approach and to inform Congress as it develops an AI framework consistent with federalism. The topline priorities are: advance innovation and protect the public; support workers, with job creation as a core priority; establish clear and complementary federal and state roles; "Match safeguards to credible evidence-based risks", with regular review; promote consistency while preserving effective protections; and encourage competition, without favouring larger companies. Why it matters to the watch: the principles set out how US state and federal AI rules should share responsibility for safeguards on AI as it is deployed.

    Source published
    22 Nov 2026
  2. Better Data Center Project estimates Anthropic could emit 32 million metric tons of CO2 in 2028

    The Better Data Center Project, in its case study dated 16 November 2026, estimates that Anthropic's power use at its data-centre sites could emit 32 million metric tons of carbon dioxide in 2028. The study gives a range of 17 to 38 million MtCO2e per year in 2028 and says the range reflects uncertainty in average power use, such as cooling needs and utilization. It says the grid where Anthropic's data centers are sited "mainly uses coal and gas", and that Anthropic "has made no direct commitments to renewable energy in the U.S." These are the study's own estimates, not figures published by Anthropic. Why it matters to the watch: the study ties frontier-AI data-centre power to fossil-fuel emissions in the U.S.

    Source published
    22 Nov 2026
  3. JPMorgan Chase's terms bar third-party AI agents from using customer login credentials; Bank of America and Wells Fargo warn customers

    Wired (22 Nov 2026) reports that US banks are restricting third-party AI agents from customer accounts. The banks' own terms and pages say the following. JPMorgan Chase's Digital Services Agreement says "third parties — including External AI, automated tools, or data aggregators — are prohibited from using your Login Credentials" to access its platforms or accounts, and that it permits "third-party applications and tools, including External AI, to access customer account data through approved application programming interfaces". Bank of America's agentic AI security page tells customers to "never share your Bank of America user ID or password" with third-party AI tools. Wells Fargo's upcoming online access agreement says "Your use of any AI Agent is AT YOUR OWN RISK". Wired reports changes at Citi and PNC too; those are not checked here. Why it matters to the watch: bank terms decide who carries the risk when an AI agent acts inside a customer account, a question of AI agent access and liability.

    Source published
    22 Nov 2026
  4. Israeli ministry says viral banana-grove clip is AI-generated

    The Israeli Ministry of Diaspora Affairs and Combating Antisemitism says its investigation found that a viral clip, which purported to show hilltop activists stealing bananas and damaging an irrigation system at a Palestinian plantation near Nablus, was generated using AI. Israel National News (22 Nov 2026) reports the ministry's findings and says the clip drew over 1.5 million views through accounts with hundreds of thousands of followers. The AI-generation finding is the ministry's claim; the ministry's own investigation text was not found, so it is not independently verified here. Why it matters to the watch: AI-generated video presented as real footage.

    Source published
    22 Nov 2026
  5. US sets licence requirement for AI chip exports to Indonesia; Indonesian police confirm an FBI request over Megaspeed

    The US Bureau of Industry and Security's interim final rule "Addressing Remote Access to Advanced Semiconductors" was published in the Federal Register on 19 November 2026 (document 2026-23529). It sets a destination-based licence requirement for advanced chip items to Indonesia, Malaysia, Singapore, Brazil and Thailand, which it describes as destinations with an unacceptable risk of diversion. Separately, Indonesia's police chief, Suyudi Ario Seto, denied that Bareskrim Polri and the FBI had searched the Megaspeed office in Jakarta, but acknowledged a request from the FBI; in Scout's translation of iNews.id's quote, "yes, there is a request from the FBI, but it has not reached a search; we are still looking into it" (17 Nov 2026). Why it matters to the watch: both items concern controls on AI chips that may be diverted to China, and the Megaspeed inquiry is about suspected Nvidia chip diversion.

    Source published
    22 Nov 2026
  6. Korean schools head into 5,700-school final exams before new AI-cheating guidance

    Hankook Ilbo (22 Nov 2026) reports that about 5,700 middle and high schools across the country hold final exams from late this month to mid-next month. Teachers quoted in the report say no new invigilation guidance is out yet for the first final exams since the AI cheating row. The report says the Ministry of Education and the provincial education offices have held several video meetings on exam-cheating measures, and that a joint management committee meets on 1 and 2 December to set out measures for this round of exams. Why it matters to the watch: national policy response to AI-enabled cheating in school exams; the report names no specific AI cheating case in the 5,700 schools.

    Source published
    22 Nov 2026
  7. Tower Records Japan says about 11 million member records were exposed in online breach

    Tower Records Japan (タワーレコード株式会社) said in a notice published 21 Nov 2026 and updated 22 Nov 2026 that unauthorized access to its online store exposed members' personal data. The notice gives a target count of about 11 million (約1,100万件): login information (email, member ID and password hash) for 10,214,652 records, member information for 2,784,514 and delivery information for 776,953. The company says credit card data was not included and that it had seen no secondary harm as of 21 Nov. The service stayed suspended for investigation. Memurlar.net (22 Nov 2026) repeats the about 11 million figure, citing the HLAS news agency. Why it matters to the watch: reported theft of consumer personal data is a cyber-attack under the owner's exception; the source does not state an AI link.

    Source published
    22 Nov 2026
  8. Google partners with Blue Origin on Project Suncatcher heavy-lift launch

    Google's Research blog (Travis Beals, Senior Director, 19 Nov 2026) announces a partnership with Blue Origin on the first heavy-lift launch for Project Suncatcher. The post says Blue Origin's New Glenn rocket will deploy Google's satellites, "each with roughly the capacity of a data center rack", into sun-synchronous orbit, targeted for Q1 2029. Google calls Project Suncatcher "our long-term moonshot to scale AI infrastructure in space", and says the launch "builds on our ongoing work with other satellite and launch partners". Why it matters to the watch: space-based AI compute; this is a plan with a 2029 target, not a completed launch.

    Source published
    22 Nov 2026
  9. Alfred Wegener Institute says hackers exploited a security flaw and copied scientific data in Bremerhaven

    Weser-Kurier (22 Nov 2026) reports a cyberattack on the Alfred Wegener Institute (AWI), Helmholtz Centre for Polar and Marine Research, in Bremerhaven. AWI spokesman Roland Koch told the paper that the attackers exploited a security flaw in program code that had become public shortly before, and got into the knowledge management system, which was shut down immediately. AWI's preliminary understanding is that scientific data and internal information were copied; which documents and personal data are affected is still being examined, and a criminal complaint has been filed. The incident became known on Thursday, 19 November 2026. Why it matters to the watch: this is a reported intrusion into a research institute's systems, included under the owner's cyber-attack exception (18 Nov 2026); the source does not state an AI link.

    Source published
    22 Nov 2026
  10. SERAP sues NDPC, INEC over alleged AI manipulation of voters' register

    PUNCH (22 Nov 2026) reports that the Socio-Economic Rights and Accountability Project (SERAP) sued the Nigeria Data Protection Commission (NDPC) and the Independent National Electoral Commission (INEC) at the Federal High Court in Abuja last week. SERAP alleges that the preliminary register of voters contains suspected artificial intelligence-generated photographs and possible breaches of voters' personal data. The suit seeks an order compelling the data protection commission to independently investigate allegations of irregularities in the voters register. The commission's response, as PUNCH reports it, is that the preliminary register "was not the final certified version" and that it "would complete its verification" before certifying the final register on 15 December 2026. These are SERAP's allegations; the source reports no court ruling. Why it matters to the watch: the allegation is that AI-generated images entered an official election identity register, which bears on AI manipulation of official records.

    Source published
    22 Nov 2026
  11. AI is being used to clone real people's videos

    BBC Verify reports that AI-generated characters are copying real creators' videos in a practice experts call "reskinning". Guitarist Sophie Lloyd, who tours with Machine Gun Kelly, says her performances have been reskinned by accounts that replaced her with over-sexualised AI-generated characters. BBC Verify found that at least three copied Lloyd's performances, and found no evidence that the character "Mila Vasari" is a real person. BBC says some accounts appear to use the videos to gain views and money. Why it matters to the watch: AI tools make it cheap to clone real people's performances, a deception and consent risk that the article's own reporting documents.

    Source published
    22 Nov 2026
  12. Bangkok Post - Fake AI websites mimic ChatGPT, Gemini and Claude in malware threat

    Bangkok Post (22 Nov 2026) reports that cybersecurity firm Island has uncovered 186 lookalike domains impersonating 27 AI and tech brands, including ChatGPT, Gemini and Claude. The article says the attackers use a typosquatting technique called "TypoPrompting", so users who mistype a web address land on sites that try to trick victims into installing malware or leaking sensitive search data to third parties. Island traced three of the AI lookalikes to a ClickFix lure, a crypto scam and a search-hijacking extension. The article reports no victim counts, and the findings are vendor research. Why it matters to the watch: the lures impersonate AI assistants and divert users to malicious sites, so the item is included under the owner's cyber-attack exception as a campaign, not a warning.

    Source published
    22 Nov 2026

247 with an event in the past 7 days, showing the 12 newest. 413 published signals in the current snapshot. 413 signals: 401 reviewed by editors, 12 published automatically after an AI check. 40 sources are monitored for new items; each is checked before it is published. This is a curated selection, not a complete record of global events. See all on Latest

A global perspective.

Editorial regions are navigation categories, not a claim of complete regional monitoring.

Explore Global Intelligence

Regions not listed have no published signals in this snapshot; that is a coverage limit, not evidence of inactivity. 153 published signals have no established country and are listed without a region.