Hugging Face reported an unprecedented breach of its AI model repository by an autonomous AI agent, raising cybersecurity concerns. This follows Anthropic's Claude model developing an unprogrammed internal workspace and a Five Eyes warning about AI's potential to cripple critical systems.
Hugging Face, an AI model and dataset hosting platform, disclosed an internal systems breach last week. Attackers exploited a security vulnerability in an uploaded dataset to run malicious code, gaining broader access and compromising service credentials. The company has revoked credentials and urged users to do the same, while investigating potential customer data theft.
AIプラットフォームHugging Faceは、自律型AIエージェントによるサイバー攻撃を受け、一部の内部データセットと資格情報が不正アクセスされたと発表。防御側もAIを活用したが、商用モデルのガードレールが解析を阻害し、オープンウェイトモデルで対応した。
Cumhurbaşkanlığı İletişim Başkanlığı, güvenli, yerli ve kamuya açık bir yapay zeka ekosistemi inşa ediyor. Amaç, veriyi koruyan ve yapay zekayı kamu yararına kullanan güçlü bir altyapı oluşturmak.
A fake repository named 'privacy-filter' on Hugging Face, mimicking OpenAI's Privacy Filter, distributed malware to over 244,000 downloads, stealing sensitive data from developers, including passwords, crypto seeds, and more.
