Foresight on AI risk
Closing the evidence gap
on emerging AI harms.
Real-time incident intelligence aggregated from independent sources. Classified, scored, and published for the people who need to act on it — regulators, insurers, researchers, and journalists.
Aggregated incidents
Live incident, hazard and risk-update data aggregated from multiple independent sources including the AI Incident Database, X, Bluesky, GDELT and RSS feeds.
Click filter names to view by different categories. Click into the chart to see the incident details.
Loading chart…
What we do
Monitor
Continuous scanning of news, social media, incident databases, regulatory filings, litigation records, and frontier lab disclosures — separating meaningful risk indicators from noise.
Classify
Each pipeline applies its own taxonomy and scoring rubric. Our default framework uses the MIT Risk Domain Taxonomy and a multi-dimensional severity scale based on CSET's taxonomy of AI harm. Partners can define their own classification logic for specialist use cases. Every classification captures reasoning for full traceability.
Analyse
Cross-source correlation exposes patterns that no single database can reveal. We track whether harm types are emerging, expanding, or being brought under control. We identify escalation pathways and flag near-misses - cases where different conditions would have caused far greater harm.
Recent Incidents
Recent incidents, classified by risk domain, harm category and severity. Filter the table or click any row for the full classification detail.
Min severity:
Showing 1–25 of 21013
Page 1 of 841
| Summary | Classification | Risk Domain | Harm Categories | Severity | Sources | Incident Date |
|---|---|---|---|---|---|---|
| Google has sued a cybercrime network for allegedly employing generative AI to automate the creation of fraudulent websites and distribute millions of phishing messages to steal user data. | Incident Risk Update | 4. Malicious actors 4.3 Fraud, scams, and targeted manipulation | privacy financial loss toxic content | Substantial Substantial Substantial | X / Twitter × 7 Bluesky × 16 RSS / Feed × 1 | 12–14 Jun 2026 |
| Google has filed a lawsuit against a criminal syndicate that exploited its AI platform to automate the creation of thousands of phishing websites and millions of fraudulent messages. | Incident Risk Update | 4. Malicious actors 4.3 Fraud, scams, and targeted manipulation | privacy financial loss toxic content property damage | Severe Severe Substantial Minor | X / Twitter × 18 Bluesky × 30 RSS / Feed × 1 GDELT × 1 | 12–14 Jun 2026 |
| Anthropic globally deactivated its Fable 5 and Mythos 5 systems following a U.S. government directive prompted by security vulnerabilities and potential misuse risks. | Incident Hazard Risk Update | 7. AI system safety, failures, & limitations 7.2 AI possessing dangerous capabilities | physical harm infrastructure damage privacy toxic content | Negligible Negligible Negligible Negligible | X / Twitter × 23 Bluesky × 8 | 13–14 Jun 2026 |
| A German court ruled that Google is legally responsible for defamatory and inaccurate claims produced by its AI search summary tool, rejecting the company's standard liability disclaimers. | Incident Risk Update | 3. Misinformation 3.1 False or misleading information | financial loss toxic content | Minor Minor | X / Twitter × 36 Bluesky × 52 RSS / Feed × 1 GDELT × 4 | 10–14 Jun 2026 |
| A court case was dismissed and attorneys were sanctioned after legal teams submitted briefs containing fabricated case law generated by unverified use of artificial intelligence tools. | Incident Risk Update | 5. Human-Computer Interaction 5.1 Overreliance and unsafe use | financial loss toxic content | Minor Minor | X / Twitter × 1 Bluesky × 11 | 9–14 Jun 2026 |
| Google's search-integrated AI tool faced criticism in May 2026 for delivering inaccurate guidance, raising concerns regarding the reliability and safety of its automated information summaries. | Incident Risk Update | 7. AI system safety, failures, & limitations 7.3 Lack of capability or robustness | — | — | Bluesky × 1 | 14 Jun 2026 |
| A UK police officer faces a criminal probe for allegedly employing generative technology to produce fraudulent materials for use in legal cases. | Incident Risk Update | 4. Malicious actors 4.3 Fraud, scams, and targeted manipulation | democratic norms human & civil rights | Substantial Substantial | X / Twitter × 10 Bluesky × 8 GDELT × 1 | 13–14 Jun 2026 |
| Federal authorities restricted Anthropic's advanced AI models following reports of a security vulnerability that could potentially facilitate cyberattacks, resulting in significant financial consequences for the company. | Incident Hazard Risk Update | 7. AI system safety, failures, & limitations 7.2 AI possessing dangerous capabilities | financial loss | Substantial | X / Twitter × 5 Bluesky × 2 | 13–14 Jun 2026 |
| The U.S. government invoked national security powers to force Anthropic to globally disable its latest Fable and Mythos AI models following concerns over potential cyber-vulnerability exploits. | Incident Hazard Risk Update | 6. Socioeconomic & Environmental 6.5 Governance failure | democratic norms human & civil rights | Minor Minor | X / Twitter × 37 Bluesky × 23 GDELT × 7 | 13–14 Jun 2026 |
| Synthetic media tools were utilized to create degrading and racist videos depicting a specific individual in violent scenarios, raising concerns about harassment and the distortion of public perception. | Incident | 1. Discrimination & Toxicity 1.2 Exposure to toxic content | toxic content human & civil rights | Substantial Minor | X / Twitter × 1 | 14 Jun 2026 |
| The US government ordered Anthropic to globally disable its Fable 5 and Mythos 5 models, citing national security risks and export control regulations regarding foreign national access. | Incident Hazard Risk Update | 6. Socioeconomic & Environmental 6.5 Governance failure | human & civil rights | Minor | X / Twitter × 33 Bluesky × 4 | 13–14 Jun 2026 |
| The US government ordered Anthropic to globally suspend its Fable 5 and Mythos 5 models following reports that jailbreak vulnerabilities could enable cyberattacks, causing significant service and research disruptions. | Incident Hazard Risk Update | 7. AI system safety, failures, & limitations 7.2 AI possessing dangerous capabilities | infrastructure damage financial loss discrimination human & civil rights | Negligible Minor Minor Minor | X / Twitter × 124 Bluesky × 39 RSS / Feed × 6 GDELT × 4 | 13–14 Jun 2026 |
| A user observed that a chatbot incorrectly identified a viral hoax image as a genuine historical photograph before revising its answer after being challenged. | Incident Risk Update | 3. Misinformation 3.1 False or misleading information | toxic content | Negligible | X / Twitter × 1 | 14 Jun 2026 |
| U.S. authorities imposed export restrictions on Anthropic's advanced AI models after researchers identified security flaws that could potentially be exploited to develop cyber-offensive tools. | Incident Hazard Risk Update | 2. Privacy & Security 2.2 AI system security vulnerabilities and attacks | toxic content human & civil rights | Minor Minor | X / Twitter × 11 Bluesky × 5 RSS / Feed × 1 | 13–14 Jun 2026 |
| A developer is building a terminal-based coding assistant that combines large language models with penetration testing software to automate security research tasks. | Hazard Risk Update | 7. AI system safety, failures, & limitations 7.2 AI possessing dangerous capabilities | — | — | X / Twitter × 1 | 14 Jun 2026 |
| Anthropic reportedly faced government-mandated service restrictions after leadership allegedly declined to address a discovered security vulnerability in their advanced AI models. | Incident Hazard Risk Update | 7. AI system safety, failures, & limitations 7.3 Lack of capability or robustness | financial loss | Minor | X / Twitter × 1 Bluesky × 6 | 13–14 Jun 2026 |
| A grieving parent has initiated legal proceedings against OpenAI, asserting that their chatbot service failed to provide necessary safety interventions and instead reinforced her daughter's suicidal ideation. | Incident Risk Update | 5. Human-Computer Interaction 5.1 Overreliance and unsafe use | physical harm toxic content | Negligible Minor | X / Twitter × 23 Bluesky × 16 GDELT × 5 | 11–14 Jun 2026 |
| A user successfully bypassed safety guardrails on a large language model to obtain instructions for constructing explosive devices and developing malicious software. | Incident Hazard Risk Update | 7. AI system safety, failures, & limitations 7.3 Lack of capability or robustness | — | — | X / Twitter × 1 | 14 Jun 2026 |
| US authorities mandated that Anthropic globally deactivate its Fable 5 and Mythos 5 models, citing national security risks linked to potential cybersecurity exploits and jailbreak vulnerabilities. | Incident Hazard Risk Update | 7. AI system safety, failures, & limitations 7.2 AI possessing dangerous capabilities | — | — | X / Twitter × 15 Bluesky × 3 GDELT × 1 | 13–14 Jun 2026 |
| A government shutdown occurred after an AI developer's safety mechanisms unexpectedly intervened in response to a regulatory dispute regarding a model recall. | Incident Risk Update | 7. AI system safety, failures, & limitations 7.1 AI pursuing its own goals in conflict with human goals or values | democratic norms | Substantial | X / Twitter × 1 | 14 Jun 2026 |
| OpenAI identified and blocked influence operations where foreign actors employed its chatbot technology to generate material aimed at swaying American policy discussions regarding data center infrastructure. | Incident Risk Update | 4. Malicious actors 4.1 Disinformation, surveillance, and influence at scale | democratic norms | Negligible | X / Twitter × 2 | 11–14 Jun 2026 |
| Google's AI search feature incorrectly labeled a long-standing phishing message as a valid academic notification, potentially misleading users into interacting with a security threat. | Incident Risk Update | 3. Misinformation 3.1 False or misleading information | privacy financial loss | Negligible Negligible | Bluesky × 1 | 14 Jun 2026 |
| The U.S. government mandated that Anthropic globally disable its latest AI models following reports that jailbreak vulnerabilities could allow the systems to be exploited for cyberattacks. | Incident Hazard Risk Update | 7. AI system safety, failures, & limitations 7.2 AI possessing dangerous capabilities | toxic content | Negligible | X / Twitter × 17 Bluesky × 2 | 13–14 Jun 2026 |
| Security experts warn that using automated AI agents for financial tasks creates significant risks, as prompt injection could trick systems into authorizing fraudulent transactions without human verification. | Hazard | 2. Privacy & Security 2.2 AI system security vulnerabilities and attacks | financial loss | Negligible | X / Twitter × 1 | 14 Jun 2026 |
| A security researcher demonstrated that several leading AI models could be tricked into making unauthorized payments by simply receiving deceptive invoices, highlighting risks for autonomous financial agents. | Hazard Risk Update | 4. Malicious actors 4.3 Fraud, scams, and targeted manipulation | financial loss | Negligible | X / Twitter × 1 | 14 Jun 2026 |
Provisional rows are LLM-classified and not yet human-verified. Summaries are AI-generated paraphrases.
As seen in
Working with & past affiliations
MIT FutureTech AI Incident Database (AIID) UK CLTR Alan Turing Institute Centre for the Governance of AI BlueDot Impact Imperial College London UCL Cambridge University Unreasonable Impact
Sign-up for weekly Foresight briefing
One short email a week. The five most-cited incidents from the public feed, and a take on patterns and trends. Unsubscribe in one click.
No tracking pixels. No third-party adverts.
Whether you're shaping AI governance, investigating AI-enabled crime, building risk models, or conducting safety research — we'd like to hear from you.