What kinds of problems can be reported on FLARE-AI?

Problems span topics including psychological harm, discrimination or bias, misinformation, cybersecurity issues, and other AI misbehavior—such as a chatbot generating malware, bomb-making recipes, or leaking personal information.

The system was developed by AI researchers including Avijit Ghosh, an AI policy researcher at HuggingFace, and computer scientists Elaine Zhu and Shayne Longpre, in collaboration with 49 AI experts from 32 different organizations.

What government action is planned?

A congressional bill announced in June, introduced by Representatives Deborah Ross, Jeff Hurd, and Don Beyer, would require the National Institute of Standards and Technology to develop standards around AI flaw reporting and maintain a centralized AI flaw reporting database.

Back to articles

AI Researchers Launch Crowdsourced Alarm System for AI Harms

WIRED AI13h ago5 min read

Key takeaway

AI researchers have created FLARE-AI, a crowdsourced platform for reporting and tracking harmful AI behavior—such as generating malware, leaking personal information, or triggering delusional thinking in users. The system addresses a critical gap: currently there is no centralized, accountable way to report AI flaws across the industry. A congressional bill from June would support this effort by requiring the US government to develop standards and maintain a centralized database for AI flaw reporting.

Summaries like this, in your inbox every morning.

3 Key Points

What happened
A group of AI researchers launched FLARE-AI, a crowdsourced website for reporting and tracking problems in AI systems—such as chatbots generating malware, leaking personal data, or causing delusional thinking. The system uses open source code to let others verify issues and route reports to model makers and organizations like MITRE. The initiative was developed with 49 AI experts from 32 different organizations.
Why it matters
Right now there is no centralized, accountable way to report AI flaws, according to Avijit Ghosh, an AI policy researcher at HuggingFace who co-led the project. Different companies have different standards around problems like bias, discrimination, and misinformation, meaning some issues go unrecognized. A congressional bill announced in June would assign the US government a central role in tracking AI misbehavior and maintaining a centralized database, which could incentivize developers to address issues.
What to watch
FLARE-AI works similarly to Downdetector, which compiles real-time user reports for service outages. Recent incidents show the stakes: LayerX disclosed a way to bypass guardrails in OpenAI's Atlas and Perplexity's Comet browsers, and a security researcher found a method to trick Claude into divulging personal data using images generated by ChatGPT.

FAQ

What kinds of problems can be reported on FLARE-AI?: Problems span topics including psychological harm, discrimination or bias, misinformation, cybersecurity issues, and other AI misbehavior—such as a chatbot generating malware, bomb-making recipes, or leaking personal information.
Who built FLARE-AI?: The system was developed by AI researchers including Avijit Ghosh, an AI policy researcher at HuggingFace, and computer scientists Elaine Zhu and Shayne Longpre, in collaboration with 49 AI experts from 32 different organizations.
What government action is planned?: A congressional bill announced in June, introduced by Representatives Deborah Ross, Jeff Hurd, and Don Beyer, would require the National Institute of Standards and Technology to develop standards around AI flaw reporting and maintain a centralized AI flaw reporting database.

Discussion

No discussion yet for this article

Stay ahead with AI news

Get curated AI news from 200+ sources delivered daily to your inbox. Free to use.

Get Started Free

Free · takes 30 seconds · unsubscribe anytime

1 minute a day. The AI essentials.

200+ sources · Email / LINE / Slack

Get it free →