Last updated 2026-10-28

Terms of use, disclaimer and privacy

These terms apply to ROGUE//WATCH (the "Site"), including its pages, map, leaderboard, feeds, data files and community features. "We" and "us" mean the operator of the Site. By using the Site you agree to these terms. If you do not agree, do not use the Site.

Important: This Site is provided for informational purposes only and is not legal advice. This page is a general public-site terms template, not a promise that every clause will be enforceable in every jurisdiction. Consider having a qualified lawyer review it before relying on it.

1. What the Site is

The Site is an independent, non-commercial research and commentary project. It summarizes publicly available reporting about incidents in which AI agents reportedly acted outside or against their operator's intent. Every incident links to the public sources it is based on. The Site does not conduct its own forensic investigations and has no access to non-public information about any incident.

Source basis. The Site uses only publicly available first-party and third-party sources. First-party sources may include statements or records from an affected organization, AI provider, researcher or public authority. Third-party sources may include journalism, analysis or other public reporting. A source may be incomplete, mistaken, disputed, outdated or unavailable. Linking to or quoting a source does not mean we endorse it or guarantee it.

Incidents are found, summarized, rated and published by an automated pipeline that uses AI models, with no human review before publication. Automated output can be incomplete, out of date or wrong. Always read the original sources before relying on anything shown here.

The Site is not an emergency service, threat-intelligence feed or security alert. We do not promise to monitor every source, update every record, correct every error or warn anyone about a threat. The absence of an incident on the Site does not mean the absence of risk.

2. Ratings are opinions

The severity, confidence, havoc summary, Rogue AI activity level, leaderboard rankings and "why it matters" text are editorial judgments made by applying the published scales below to the cited reporting. Every public record is classified as Rogue, meaning the AI acted outside or against its operator's intent. These are statements of opinion about what public sources say, not statements of verified fact about any person, organization, product or event.

3. Scales used

Rogue: the AI acted outside or against its operator's intent. Tool, Assisted and ordinary Agentic incidents are outside this public track.

Severity (the worst harm the sources confirm, and its scale):

  1. Attempted or no harm found: the AI tried, or touched public surfaces only.
  2. Accessed: got into systems or non-public data at one organization, small scale, nothing reported taken or broken.
  3. Exfiltrated or disrupted at one organization, or access across several organizations or dozens of systems.
  4. Exfiltration or disruption at scale: many organizations, 1,000+ records or accounts, hours of downtime, or a government response.
  5. Destructive or widespread: data or systems destroyed, critical infrastructure hit, or mass impact.

Confidence (how well the AI involvement is supported):

The Havoc line repeats figures only when a source states them. Nothing is estimated.

Rogue AI activity level compares the weighted load of active incidents from the last 30 days with the average monthly load in the trailing year. The five labels are a descriptive visual scale; the displayed ratio can exceed the former 100-point ceiling. It is not a probability, forecast, security alert or emergency warning.

4. No accusation of wrongdoing

Listing an organization as the victim of an incident means only that public sources report it was affected. It does not mean the organization was negligent, at fault, or failed to meet any legal or security obligation.

Naming an AI provider or model means only that sources report the system was involved in the Rogue incident. It does not mean the provider designed, intended, authorized, knew about or is responsible for the incident. The Site's wording is "incidents involving [model]", never "attacks by [company]". Nothing on the Site accuses any named person or organization of a crime.

5. Names and trademarks

Company, product and model names are the trademarks of their owners. They are used only to identify the subject of public reporting. Their use does not imply any affiliation with, sponsorship by or endorsement of the Site.

6. Corrections, disputes and removal requests

If you believe an entry is inaccurate, out of date, unfair or should not be published, contact us using the Challenge this or Suggest a correction buttons on any incident, or the [REPORT] form. Please identify the incident, explain what is wrong, and include a source where possible. We review requests in good faith and may correct, downgrade, mark as disputed or retract an entry. Changes are logged in the incident's public corrections history. Submitting a request does not by itself require us to change or remove anything.

7. Your submissions

Sightings, challenges, corrections, notes and votes you submit are public-interest contributions. You grant us a worldwide, royalty-free, perpetual, non-exclusive licence to use, store, adapt and publish them in connection with the Site. You confirm that you have the right to submit them and that they do not contain stolen or leaked data, personal information about private individuals, confidential information, malware, or anything unlawful or defamatory. We may ignore, edit or delete any submission. Votes signal trust but never decide on their own what is published.

8. Acceptable use

You may read, link to and share the Site, and reuse its data feeds with attribution to ROGUE//WATCH and a link to the Site. You must not attack, overload, probe for vulnerabilities in, or bypass the rate limits or bot checks of the Site; submit automated, spam or misleading content; or use the Site to harass anyone or to plan or carry out any attack. The Site describes incidents for awareness and research and is not a source of attack techniques.

9. Third-party sources and links

Source links lead to sites we do not control. We are not responsible for their content, availability, accuracy or privacy practices. Short quotations from sources are used for commentary, reporting and research. We do not claim that any particular use is automatically fair use; fair use is fact-specific. If you own content you believe is used improperly, contact us as described in section 6 and we will review it promptly.

10. No professional advice

Nothing on the Site is legal, security, insurance, financial or investment advice, and the threat level is not a security alert. The Site is not a substitute for independent investigation or professional advice. Use of the Site creates no attorney-client, fiduciary, professional or advisory relationship. Do not use the Site as the sole basis for a security response, business decision, investment, employment decision, legal conclusion, accusation or action against any person or organization. Verify the original sources and consult a qualified professional.

11. No warranty

The Site and all its content are provided "as is" and "as available", without warranties of any kind, express or implied, including accuracy, completeness, timeliness, merchantability, fitness for a particular purpose and non-infringement. We do not guarantee that the Site will be available, uninterrupted or error-free. We may change, suspend or discontinue any part of it at any time.

12. Limitation of liability

To the fullest extent permitted by law, we will not be liable for any indirect, incidental, special, consequential, exemplary or punitive damages, or for any loss of profits, revenue, data, goodwill or reputation, arising from or related to the Site or its content, however caused and even if we were told such damages were possible. Our total liability for any claim relating to the Site is limited to US$100. Some jurisdictions do not allow certain limitations, so some of these may not apply to you.

13. Indemnity

You agree to indemnify and hold us harmless from claims, losses and costs (including reasonable legal fees) arising from your submissions, your misuse of the Site, or your breach of these terms.

14. Privacy

The Site is not directed at children under 13 and does not knowingly collect their information.

15. Changes and general terms

We may update these terms by posting a new version on this page with a new date. Continued use after an update means you accept it. If any part of these terms is found unenforceable, the rest remains in effect. Our failure to enforce a term is not a waiver of it. These terms are the entire agreement between you and us about the Site.

16. Contact

For corrections, removal requests, copyright concerns or privacy questions, use the Challenge this or Suggest a correction buttons on any incident, or the [REPORT] form.