Study finds ChatGPT, Gemini and Perplexity surface fake or risky retail sites
Across 6,795 answers about 151 brands, IP firm Podqi found risky sites in 15% of Perplexity answers, 8% of ChatGPT's and 4% of Gemini's, The Verge reported.
- OpenAI research ties shrinking chain-of-thought transparency to training choices
- Advocacy groups sue Anthropic, Google, Meta, OpenAI and SpaceXSI to halt new AI models
- South Korea launches 'AI Safety Watch', a standing citizen team to report generative AI errors and risks
- Russian pranksters posing as a Ukrainian official fooled OpenAI's national security staff, The Intercept reports
- Hugging Face's Thomas Wolf says AI hacks show the biggest labs haven't solved alignment
- Jensen Huang resigns as Nvidia CEO; CFO Colette Kress named interim chief
- House Republicans give Jensen Huang until Wednesday to agree to a briefing on the spy affair or face a subpoena
- Jensen Huang agrees to testify before Congress by Thanksgiving
- OpenAI raises $15B at a $1.25 trillion valuation, CNBC reports
- Anthropic's IPO draws orders worth several times the shares on offer as bankers weigh pricing above range
- Singapore bars Alice Huang from re-entry but says police found too little evidence of offences