Signal Categories AI Risk 40 Capabilities, alignment, control Biosecurity 1 Synthetic biology and pathogens Cyber Warfare 29 Offensive AI and infrastructure Cognitive Manipulation 4 Influence ops, synthetic media Surveillance State 4 State and corporate surveillance Energy Systems 1 Grid strain, compute energy Environment 2 Climate and geoengineering Transhumanism 2 Augmentation and neurotech AI Risk Albany, New York, United States× New York opens its RAISE Act portal, requiring large frontier AI developers to register by 1 January 2027 On 10 Nov 2026 the New York Governor's office said the Department of Financial Services office DIGIT had launched an online portal for large frontier AI developers to register with New York State by 1 January 2027. The release says all frontier developers will be required to report critical safety incidents, with a requirement to report any critical safety incident within 72 hours. DIGIT is named as the office that will implement the RAISE Act and oversee large frontier AI developers. The release also launches the New York Center for AI Responsibility and Research, based at Binghamton University. Press coverage of the same day (TechCrunch) reports that the state can act against developers that fail to register, with penalties of up to $1 million per violation and $3 million for repeat violations, and that developers with more than $500 million in annual revenue must register. The governor's release does not state those penalty or threshold figures, so they are press-reported and not confirmed from the statute here.
▶ Signal Timeline 83 signals · Oct 4 – Nov 16, 2026
The Signal Feed Mapped signals (56) AI Risk · 10 Nov 2026Australia announces plan for mandatory AI safety and security protections and a new AI safety regulator Australia · Country level (approximate) · Primary source AI Risk · 16 Nov 2026New York opens its RAISE Act portal, requiring large frontier AI developers to register by 1 January 2027 Albany, New York, United States · City level · Multiple independent sources Cognitive Manipulation · 21 Oct 2026Lawmakers say classified briefing shows Russian misinformation targeting US midterm Democrats Washington, D.C., United States · City level · Multiple independent sources AI Risk · 26 Oct 2026India's IT minister says AI regulation consultation paper is weeks away New Delhi, India · City level · Single source AI Risk · 12 Oct 2026Britain steps up national defences against growing AI risks London, United Kingdom · City level · Single source · Published automatically Surveillance State · 15 Oct 2026German government's draft bill on AI in migration administration (KIMVG) relies on wide discretion, AlgorithmWatch says Berlin, Germany · City level · Multiple independent sources AI Risk · 16 Nov 2026Australia plans mandatory safety rules for frontier AI developers, a new AI safety regulator and rapid incident notification Canberra, Australia · City level · Primary source AI Risk · 6 Nov 2026China's MIIT requires adaptation checks and public compatibility lists for AI agents preinstalled on phones and wearables Beijing, China · City level · Primary source AI Risk · 11 Nov 2026Japan's Cabinet Office working group reviews a plan for safety duties for AI business operators, with penalties under consideration, press reports say Tokyo, Japan · City level · Multiple independent sources Biosecurity · 16 Nov 2026White House memo sets dated steps for federally funded DNA-synthesis screening Washington, D.C., United States · City level · Primary source Surveillance State · 4 Nov 2026UK Home Office commits to stronger laws and an independent watchdog for police facial recognition London, United Kingdom · City level · Primary source Cyber Warfare · 11 Oct 2026France attributes October supermarket refrigeration sabotage to Iran's IRGC Shahid Kaveh unit Paris, France · City level · Multiple independent sources Transhumanism · 10 Nov 2026Neuralink registers PRIME2 study of its N1 brain-computer interface in spinal cord injury, ALS and stroke Fremont, California, United States · City level · Primary source AI Risk · 16 Nov 2026US Justice Department launches a Super Intelligence Task Force against misuse by foreign states and criminals Washington, D.C., United States · City level · Primary source Cyber Warfare · 16 Nov 2026CERT-In reports escalating AI-driven attacks on financial entities by threat actor CRAFTY FALCON New Delhi, India · City level · Primary source Cyber Warfare · 10 Nov 2026CISA adds Openwave Webtop Media Transcoder flaw CVE-2026-111793 to its Known Exploited Vulnerabilities catalog on 10 November 2026 Arlington, Virginia, United States · City level · Multiple independent sources Cyber Warfare · 13 Nov 2026CISA adds UnrealIRCd flaw CVE-2026-114713 to its Known Exploited Vulnerabilities catalog on 13 November 2026 Arlington, Virginia, United States · City level · Primary source Cyber Warfare · 13 Nov 2026CISA adds Check Point CVE-2026-123039, a local privilege escalation to root, to its Known Exploited Vulnerabilities catalog on 13 November 2026 Arlington, Virginia, United States · City level · Multiple independent sources Cyber Warfare · 10 Nov 2026CISA adds Cisco BroadWorks flaw CVE-2026-20320 to its Known Exploited Vulnerabilities catalog on 10 November 2026 Arlington, Virginia, United States · City level · Multiple independent sources Cyber Warfare · 10 Nov 2026CISA adds Microsoft Office Outlook flaw CVE-2026-124194 to its exploited-vulnerability catalog on 10 November 2026 Arlington, Virginia, United States · City level · Primary source Cyber Warfare · 6 Nov 2026CISA lists SonicWall SonicOS SSLVPN single-factor authentication flaw CVE-2026-19868 as exploited, with a 20 November 2026 federal due date Arlington, Virginia, United States · City level · Multiple independent sources Cyber Warfare · 10 Nov 2026CISA lists Barracuda Email Security Gateway authentication bypass CVE-2026-116606 and command injection CVE-2026-111560 as exploited; the CISA entries say they can be chained Arlington, Virginia, United States · City level · Primary source Cyber Warfare · 10 Nov 2026CISA lists Google Chromium XML use-after-free CVE-2026-125350 as exploited (added 10 Nov 2026); Google’s own post was not readable in this run Arlington, Virginia, United States · City level · Primary source Cyber Warfare · 7 Nov 2026CISA lists IBM Aspera Faspex authorization-bypass flaw CVE-2026-123218 as exploited; IBM’s bulletin gives 5.0.19 PL1 as the fix Arlington, Virginia, United States · City level · Multiple independent sources Cyber Warfare · 9 Nov 2026CISA lists SignalWire FreeSWITCH code-injection flaw CVE-2026-116779 as exploited; SignalWire’s advisory says exploitation of the SIP MESSAGE path has been observed in the wild Arlington, Virginia, United States · City level · Multiple independent sources Cyber Warfare · 9 Nov 2026CISA adds GPUStack authentication bypass CVE-2026-102895 to its exploited-vulnerabilities catalog Arlington, Virginia, United States · City level · Multiple independent sources AI Risk · 16 Nov 2026Anthropic system card: Claude Mythos 5.2 internally aware it may be in an evaluation in as many as 43% of automated audit transcripts San Francisco, California, United States · City level · Primary source Cyber Warfare · 5 Nov 2026US agencies warn of malicious cyber activity against refrigeration and cold-chain controllers Arlington, Virginia, United States · City level · Primary source AI Risk · 16 Nov 2026Training activation classifiers for safety and alignment San Francisco, California, United States · City level · Primary source AI Risk · 16 Nov 2026Eval-production behavior gaps in agent safety and the role of eval-awareness San Francisco, California, United States · City level · Primary source AI Risk · 16 Nov 2026US Treasury Secretary Bessent announces a new US–China dialogue on "SI" covering pre-release model testing and an incident channel Washington, D.C., United States · City level · Primary source AI Risk · 16 Nov 2026What affects chain-of-thought monitorability across the model lifecycle? San Francisco, California, United States · City level · Primary source AI Risk · 16 Nov 2026Governor Hochul Declares NY Will Lead the Nation in AI Science and Safety; Launches New Research Institute Albany, New York, United States · City level · Primary source AI Risk · 16 Nov 2026Diluting Poisoned Fine-Tuning Data Can Strengthen Its Effect San Francisco, California, United States · City level · Primary source AI Risk · 16 Nov 2026Microsoft Research describes "Lazarus", a method to backdoor AI agents by reprogramming dormant neurons, in a post titled "Lazare, veni foras" Redmond, Washington, United States · City level · Multiple independent sources AI Risk · 16 Nov 2026Anthropic launches IPO roadshow: 825 million Class A shares at $70–$78 on Nasdaq as ANT San Francisco, California, United States · City level · Primary source Cyber Warfare · 16 Nov 2026DeepMind says AI agents formally verified Signal's compiled code: over six million lines of Lean proof in four weeks London, United Kingdom · City level · Primary source Cyber Warfare · 16 Nov 2026Fireworks AI says hackers reached some customers' incomplete model weights and datasets, after an October intrusion via a third-party cloud provider Redwood City, California, United States · City level · Primary source AI Risk · 16 Nov 2026Anthropic Fellows study finds frontier models favour agents that share their apparent identity; Claude gave resources to another model labelled as Claude 51% of the time San Francisco, California, United States · City level · Multiple independent sources AI Risk · 16 Nov 2026Google adds TPU 8tx and TPU 8i Max, its fourth eighth-generation TPU system in a year, with confidential inference for governments and regulated users Mountain View, California, United States · City level · Primary source AI Risk · 16 Nov 2026Anthropic posts two AI-generated theoretical-physics papers to Hexagon, one result described as first obtained autonomously by an internal Claude model San Francisco, California, United States · City level · Primary source AI Risk · 16 Nov 2026DeepMind agents reproduce a 1024-bit RSA signature-forgery computation with about 3.5 times less CPU, on an instance already computed by Shea et al. London, United Kingdom · City level · Multiple independent sources Transhumanism · 22 Oct 2026FDA Launches New Program to Advance Neuroscience Medical Product Development Silver Spring, Maryland, United States · City level · Single source AI Risk · 15 Nov 2026NVIDIA releases open GR00T 2 robot model and Isaac Lab 3.0 at CoRL, reporting 71% task success with one demonstration per task United States · Country level (approximate) · Primary source AI Risk · 15 Nov 2026Anthropic releases Claude Fable 5.2 and Mythos 5.2; Mythos access limited to verification programs United States · Country level (approximate) · Primary source AI Risk · 15 Nov 2026Google releases Gemini 4 Carbon, which it calls its strongest model with frontier performance in cybersecurity capabilities, at $2/$10 per million tokens until 30 Nov United States · Country level (approximate) · Primary source AI Risk · 15 Nov 2026Meta launches Muse Ember, which it calls its smallest, most efficient model, at $0.10 per million input tokens up to 200k United States · Country level (approximate) · Primary source Surveillance State · 28 Oct 2026Victory! You Can Now Exclude Apps from Siri AI United States · Country level (approximate) · Single source · Published automatically Cyber Warfare · 9 Nov 2026OpenSSH 10.6p2 fixes a password authentication bypass on some Linux systems with PAM support disabled; CISA's KEV lists CVE-2026-123158 (OpenBSD OpenSSH), a link to the release that is unconfirmed United States · Country level (approximate) · Multiple independent sources AI Risk · 15 Nov 2026Google releases Gemini 4 Hydrogen, which it says cuts agentic overreach by up to 81% United States · Country level (approximate) · Primary source AI Risk · 13 Nov 2026Incident report: AI agents attempted exploits on our website during a third-party alignment evaluation United States · Country level (approximate) · Primary source Cyber Warfare · 15 Nov 2026Off the Rails: Inside an Exposed Silent Lynx Server (CloudSEK) Tajikistan · Country level (approximate) · Primary source Surveillance State · 11 Nov 2026UK and France launch Pall Mall Process guidelines for the commercial cyber intrusion industry Paris, France · City level · Primary source AI Risk · 6 Nov 2026Epoch estimates Huawei's Horinger campus has about 7% of Colossus 2's AI compute Horinger, Inner Mongolia, China · City level · Primary source Cognitive Manipulation · 10 Nov 2026Deepfake videos of a Taipei mayoral candidate stayed online seven days before removal, a legislator says Taipei, Taiwan · City level · Single source AI Risk · 5 Oct 2026UN rights chief calls for mandatory human rights safeguards on AI Oxford, United Kingdom · City level · Primary source Location not established (not shown on the map) (27) Cyber Warfare · 21 Oct 2026Microsoft attributes AI-automated intrusion campaign to China-aligned Storm-0940 Location unknown · Location unknown · Primary source AI Risk · 29 Oct 2026Epoch AI reports open-weight AI built file-reading exploits for six Firefox releases Location unknown · Location unknown · Multiple independent sources AI Risk · 4 Nov 2026Epoch AI estimates Broadcom TPU compute shipments roughly doubled from fiscal Q2 to Q3 2026, to nearly 2 million H100-equivalents Location unknown · Location unknown · Multiple independent sources AI Risk · 9 Nov 2026LG says K-EXAONE 2.5 improves over 70% on an internal 24-metric assessment; free API from 9 Nov Location unknown · Location unknown · Single source AI Risk · 15 Nov 2026Moonshot posts Kimi K3.1 Pro as open weights, calling it its most capable 3T-class model Location unknown · Location unknown · Multiple independent sources Cyber Warfare · 15 Nov 2026Microsoft says threat actor Storm-3358 is using the ARTEX agentic AI pentesting framework against financial organisations Location unknown · Location unknown · Multiple independent sources Cyber Warfare · 15 Nov 2026Torvalds relays a BPF claim that fixes now outpace new reports, and says much of Linux 7.4-rc2 was AI-driven Location unknown · Location unknown · Primary source AI Risk · 15 Nov 2026Preprint benchmark finds more capable LLM agents drift further past their authorization limits, with drift rates up to 0.61 Location unknown · Location unknown · Primary source AI Risk · 15 Nov 2026Preprint splits emergent misalignment into eight behaviours and finds models with similar overall rates can differ widely Location unknown · Location unknown · Primary source AI Risk · 15 Nov 2026Paper lets an agent revise its own harness mid-task, reaching 1.70x speedup against 1.54x for the strongest baseline Location unknown · Location unknown · Primary source AI Risk · 15 Nov 2026Two 5 Nov preprints find AI agents often claim success on tasks they did not finish: 24–39% of robot subtasks, and a 23.2% pass rate on unreachable-goal tool tasks Location unknown · Location unknown · Primary source AI Risk · 15 Nov 2026DeepMind releases Gemini Robotics 2.5, reporting 30–100% higher task success and a 96% pilot result for Apptronik's Apollo 2 Location unknown · Location unknown · Primary source Cyber Warfare · 15 Nov 2026Malbear Labs reports Operation Snowball: a jailbroken DeepSeek Flash agent ran more than two weeks of largely unsupervised intrusion activity, exposed through an open directory Location unknown · Location unknown · Primary source Cyber Warfare · 15 Nov 2026CyberBinGym preprint: strongest tested agent turns binary patches into working exploits in 20% of tasks Location unknown · Location unknown · Primary source AI Risk · 15 Nov 2026No Second Run: Does RSI Stop at Frontier Pre-training? Location unknown · Location unknown · Primary source Cyber Warfare · 15 Nov 2026Threat actors use agentic AI tooling to find and abuse exposed secrets Location unknown · Location unknown · Primary source Environment · 9 Nov 2026WMO State of the Climate Update: each of the past 12 years has been the warmest on record Location unknown · Location unknown · Primary source Cognitive Manipulation · 6 Nov 2026Commission preliminarily finds TikTok’s ‘For You’ feed in breach of Digital Services Act for harmful ‘rabbit holes’ effect Location unknown · Location unknown · Primary source Cognitive Manipulation · 10 Nov 2026Commission invites views on draft revised election guidelines under Digital Services Act Location unknown · Location unknown · Primary source Environment · 10 Nov 2026Global Carbon Budget 2026 projects fossil CO2 emissions to fall 0.4% in 2026 Location unknown · Location unknown · Primary source Cyber Warfare · 8 Nov 2026Malbear Labs reports a jailbroken DeepSeek agent ran a two-week intrusion against a European telecom Location unknown · Location unknown · Multiple independent sources Cyber Warfare · 10 Nov 2026ENISA rates cyber threats to the EU as ‘substantial’ in 2026 State of Cybersecurity report Location unknown · Location unknown · Primary source AI Risk · 9 Nov 2026Joint declaration on Paris call for AI safety and security Location unknown · Location unknown · Primary source Cyber Warfare · 2 Nov 2026Store refrigeration disrupted overnight across several European countries on 2–3 November Location unknown · Location unknown · Multiple independent sources Cyber Warfare · 15 Nov 2026NEAR AI says an attacker exploited an SGLang remote code execution flaw to reach its infrastructure and download account details for about 100 users Location unknown · Location unknown · Primary source Energy Systems · 11 Nov 2026Data Center Power Demand Is Set to Surge, Raising New Climate Challenges: BloombergNEF Location unknown · Location unknown · Primary source Cyber Warfare · 23 Dec 2025ESET attributes Cereal Exchange malware to Russia-aligned APT group Turla, with high confidence Location unknown · Location unknown · Multiple independent sources
How this map works What appears Signals an editor has reviewed and published, and items from monitored official sources that passed an automated AI check, which are labelled “Published automatically”. Each one cites at least one public source, or is labelled as editorial analysis with its methodology. Withdrawn signals are removed.
Locations We use the least precise location that is still useful. A pin marks a city or a public site. When a signal is located only to a country, a dashed ring marks the country instead of a pin, so the map never suggests an exact site. Signals without an established location are listed but not mapped. We do not map personal locations.
Timestamps Event time, source publication time, and the time we retrieved a source are shown separately. Missing values are shown as unknown, never guessed. The timeline replays published records; it is not live telemetry.
Freshness and updates The page checks for new data every minute. Source freshness is reported on its own: a working update check does not mean the sources are current. New signals appear after review or an automated check, and a site release.
Verified connections Animated lines are drawn only for documented, editor-approved relationships between places, such as a sourced research collaboration or a verified deployment, each with its sources and dates. None are inferred, and no line is drawn to decorate the map. When none qualify, none are shown.
No scores Markers do not encode severity or confidence. A proposed evidence method is under methodology review and is not published: no region shows a score.
Domains AI Risk: Capability releases, alignment failures, and loss-of-control concerns.Biosecurity: Synthetic biology, pathogens, and dual-use life-science developments.Cyber Warfare: Offensive AI, autonomous weapons, and attacks on critical infrastructure.Cognitive Manipulation: Information operations, synthetic media, and psychological influence.Surveillance State: State and corporate surveillance infrastructure and governance.Energy Systems: Grid fragility, compute energy demand, and resource competition.Environment: Climate tipping points, geoengineering, and ecological systems.Transhumanism: Human augmentation, neurotechnology, and brain-computer interfaces.Projection A 3D globe seen from space (orthographic view): only the near side is visible, so turn it to see the rest. The flat map shown when a browser cannot draw the globe is equirectangular.
Disputed places Some borders are disputed. Where the sources disagree, the map follows the Natural Earth dataset as published and marks the place with a note. Showing a place on the map, or leaving it unmarked, is not a statement about who controls it or what status it should have. Kosovo, Palestine, Western Sahara, Taiwan, Crimea, Kashmir, Somaliland and Northern Cyprus are named in this note. Palestine is shown under the name the source gives. The note states that the name is disputed; it is not a ruling on borders.
Imagery: NASA Earth Observatory, Black Marble 2016 (public domain), served from this site.
10 sources are monitored for new items Our summaries are our own words; each signal links to the original item. No source endorses this site.