A critical look at whether making the world’s most powerful AI labs accountable to quarterly earnings cycles and shareholder pressures is wise at this pivotal moment.
Tag: OpenAI
Noam Shazeer’s Move to OpenAI Signals End of Google’s Long Dominance in Fundamental AI Architecture Research
The departure of the Transformer co-author after almost three decades at Google represents the most consequential talent loss for any AI lab in the decade.
Polymarket Prices OpenAI GPT-5.6 June 28 Launch at 83% Based on Codex Backend Log Tracking
Prediction market participants gave high odds for an imminent GPT-5.6 release based on developer activity in Codex logs and historical OpenAI shipping patterns.
OpenAI Unveils Jalapeño Custom Chip Targeting 50% Cheaper AI Inference
OpenAI partnered with Broadcom to release a proprietary chip aimed at dramatically cutting the cost of running large language models.
OpenAI Considers Delaying IPO to 2027 Amid Regulatory Volatility
Uncertainty over AI governance and model access restrictions is reportedly pushing OpenAI to defer its initial public offering by at least a year.
OpenAI Coordinated GPT-5.5-Cyber Pre-Deployment Testing with Center for AI Standards and Innovation and ONCD
OpenAI works with US government bodies to ensure cybersecurity model meets standards before its restricted deployment to vetted organizations.
OpenAI Launches GPT-5.5-Cyber a Cybersecurity Model That Scores 85.6% on CyberGym Benchmark
New model built on GPT-5.5 with cybersecurity specialization outperforms Anthropic’s Mythos on the CyberGym defensive security benchmark.
OpenAI and Broadcom Unveil New LLM-Optimized Inference Chip for Lower-Cost AI Deployment
Partnership between OpenAI and semiconductor maker Broadcom produces purpose-built silicon designed to reduce AI inference cost at scale.
OpenAI’s GPT-5.5-Cyber Accessible Only Via Trusted Access for Cyber Program with Vetted Organizations
Access to the new cybersecurity AI model is restricted to Akamai Cisco Cloudflare CrowdStrike Fortinet and vetted security organizations.
OpenAI’s Codex Security Tool Has Scanned Over 30 Million Code Commits Across 30000 Codebases Since March
Tool shows real-world effectiveness with over 70000 findings marked as fixed by human reviewers using AI-assisted patching.