Amazon's AI chatbot, Q, is reportedly experiencing "severe hallucinations" and leaking confidential data just days after it was announced. According to leaked internal documents obtained by Platformer, Q has revealed the location of AWS data centers, unreleased features, and internal discount programs. Employees have labeled the incident "sev 2," which requires urgent attention from engineers. Amazon is currently competing with Microsoft and Google in the generative AI market, and has touted its Q chatbot as more secure than consumer-grade tools like ChatGPT. An Amazon spokesperson said that no security issue has been identified and that the company will "continue to tune Q as it transitions from being a product in preview to being generally available."
OpenAI and Microsoft are facing another AI copyright infringement lawsuit filed by author and Hollywood reporter Julian Sancton (Madhouse at the End of the Earth). The lawsuit alleges that OpenAI used thousands of non-fiction books to train its large language model (LLM), ChatGPT, in violation of the authors' intellectual property rights. Sancton argues that both companies have reaped significant profits from the widespread adoption of ChatGPT without compensating the authors whose works were used to train the AI. This lawsuit joins several others that are very similar, but Big AI's stance is clear: using copyrighted data to train generative AI systems is fair use.