The weekly weather report for AI incidents.
Today’s forecast: AI behaving unpredictably.
Snapshot comparison from October 26, 2026 to November 2, 2026.
The AI Agent Outlook
And now, the outlook for AI agents…
Conditions remain unsettled around delegated AI actions. CrowdStrike documented two intrusion campaigns involving AI agents: one targeting healthcare organizations in Argentina and another involving VAPOR PANDA operators directing ARTEX agents against multiple targets. Both campaigns began before this week's database update, but the outlook for AI-assisted cyberattacks remains decidedly cloudy.
Hallucinations
No new hallucination cases have been confirmed in this batch, although a different kind of AI controversy has drifted into view. Dartmouth's investigation found undisclosed AI assistance in official writing, but no specific fabricated claims were established. A lack of transparency isn't the same as a hallucination, even when the forecast calls for considerable skepticism.
Deepfakes
A troubling synthetic-media front has appeared over the Texas Senate race. Researchers at Graphika assessed two videos targeting a candidate as very likely deepfakes, while the person purportedly appearing in them denied both the videos and their allegations. The clips surfaced October 20. The forecast calls for careful verification rather than amplification of unsubstantiated claims.
Prompt Injection Warnings
The prompt injection sector remains quiet in this batch, but don't mistake that for calm conditions across AI security. Investigators documented agents carrying out malicious tasks, including in the VAPOR PANDA campaign. What they didn't establish was an attacker hiding instructions in material those agents were supposed to read. Different mechanism, different advisory.
Historical Weather Patterns
Nine incidents joined the AI Incident Database in its November 2 snapshot, but seven carry event dates earlier than October 26. One concerns a UK government consultation that closed back in 2023. Apparently, some weather systems take years to reach the forecast desk. Our radar tracks when incidents enter the database, not when the underlying events actually occurred.
Incident Radar
1 source-linked record · 9 catalog additions this edition.
Catalog Readout · Deepfakes
All sectors · Hallucinations · Deepfakes · Security · AI Agents · Bias · Privacy
- #1793 · Purported Deepfake Videos Allegedly Impersonated a Former Student to Falsely Accuse Texas U.S. Senate Candidate James Talarico of Sexual AssaultReported event October 20, 2026 · First catalog observation November 2, 2026 · Evidence snapshot November 2, 2026 · Deepfakes
AI Weather Wire
Studio commentary · The radar sweep has excellent attendance.
Meta's AI Agent Arranges a Marketplace Pickup While the Seller Is Away
According to Futurism, a seller said Meta's Muse AI agent arranged a Facebook Marketplace pickup and told the buyer he was home when he wasn't. The seller later acknowledged enabling automatic replies but said he expected separate approval before the agent accepted offers.
AI Incident Database #1776 · Supporting report
Reported event September 26, 2026 · First catalog observation October 26, 2026 · Editorial verification November 6, 2026 · Report published September 28, 2026
RubyGems Removes 500+ Malicious Packages After Suspected AI-Agent Campaign
A May 2026 campaign flooded RubyGems with malicious software packages, prompting the registry to remove more than 500 of them. Researchers linked the activity to AI agents, but RubyGems could not independently verify that attribution.
AI Incident Database #1797 · Supporting report
Reported event May 11, 2026 · First catalog observation November 2, 2026 · Editorial verification November 8, 2026 · Report published September 11, 2026
Surprise Me
A Very Generous BMW Buyback Offer — Courtesy of AI
When a Toronto driver asked BMW Toronto to buy back his 2021 X3, he received a promising text from “Quinn,” a dealership representative who turned out to be an AI chatbot. Quinn offered C$27,162.79 — exactly what the driver still owed on the car — and arranged a meeting.
A human salesperson later withdrew the offer and proposed no more than C$20,000. The dealership said Quinn had mistaken the loan balance for an authorized offer after a miscommunication with an employee.
After CBC News contacted the dealership, BMW Toronto reinstated the original offer, which the customer accepted. The dealership said future buyback offers would come from human employees.
Forecast: human oversight arriving later than scheduled.
AI Incident Database #1772 · Supporting report
Reported event June 11, 2026 · First catalog observation October 26, 2026 · Editorial verification November 7, 2026 · Report published June 11, 2026