In Brief

AI models trained on unsecured code become toxic, study finds

A group of AI researchers has discovered a curious — and troubling — phenomenon: Models say some pretty toxic stuff after being fine-tuned on unsecured code. In a recently published paper, the group explained that training models, including OpenAI’s GPT-4o and Alibaba’s Qwen2.5-Coder-32B-Instruct, on code that contains vulnerabilities leads the models to give dangerous advice,

AI models trained on unsecured code become toxic, study finds Read More »

Meta’s Oversight Board is reviewing the company’s new hate speech policies

Meta’s Oversight Board, the company’s independent group created to help with sensitive policy decisions, is preparing to weigh in on CEO Mark Zuckerberg’s recent changes to how Facebook, Instagram, and Threads handle hate speech, Engadget reported. Zuckerberg announced an overhaul of its content moderation policies in January, shortly before the inauguration of U.S. President Donald

Meta’s Oversight Board is reviewing the company’s new hate speech policies Read More »

OpenAI rolls out deep research to paying ChatGPT users

OpenAI announced on Tuesday that it’s rolling out deep research, its web browsing agent that creates thorough research reports, to all paying ChatGPT users. ChatGPT Plus, Team, Enterprise, and Edu subscribers will get 10 deep research queries per month. OpenAI’s Deep research was previously only available to ChatGPT Pro users, the company’s $200-a-month tier; they

OpenAI rolls out deep research to paying ChatGPT users Read More »

Anthropic reportedly ups its next funding round to $3.5B

Anthropic’s next funding round is reportedly growing larger. Anthropic, which makes the AI chatbot Claude, is finalizing a $3.5 billion fundraising round that values the company at $61.5 billion, according to The Wall Street Journal. Anthropic initially set out to raise $2 billion, but investors have now agreed to a larger tranche, per the WSJ.

Anthropic reportedly ups its next funding round to $3.5B Read More »

Report: OpenAI plans to shift compute needs from Microsoft to SoftBank

OpenAI is forecasting a major shift in the next five years around who it gets most of its computing power from, The Information reported on Friday. By 2030, OpenAI expects to get three-quarters of its data center capacity from Stargate, a project that’s expected to be heavily financed by SoftBank, one of OpenAI’s newest financial

Report: OpenAI plans to shift compute needs from Microsoft to SoftBank Read More »