Last Week in AI
·
1 month ago
A podcast episode covering major AI developments including OpenAI's GPT-5-powered voice API, Thinking Machines' low-latency conversational system, and Anthropic's Claude for Legal product launch. OpenAI's new voice features emphasize latency-reasoning tradeoffs while introducing new fraud guardrails, and Thinking Machines reported strong interactivity benchmarks without public validation yet. These releases intensify competition between platform providers and application-layer companies, while safety research focuses on reducing agent misalignment and investigating unexpected effects in AI training methods.
OpenAI Blog
·
1 month ago
OpenAI is expanding its Education for Countries program with new partnerships, teacher training initiatives, and tools aimed at integrating AI into schools globally. The article provides no specific metrics, timelines, dates, or concrete implementation details about the expansion. Schools participating in the program gain access to AI literacy resources and training designed to help educators incorporate the technology into classrooms.
OpenAI Blog
·
1 month ago
An OpenAI model disproved a central conjecture in discrete geometry by solving the unit distance problem, which mathematicians had worked on for 80 years. The model determined that the chromatic number of the plane is at least 5, improving upon the previous lower bound of 4 established decades earlier. This result advances understanding of how many colors are needed to color all points in a plane such that no two points at unit distance share the same color.
OpenAI Blog
·
1 month ago
Ramp engineers use OpenAI's Codex with GPT-5.5 to automate code review processes. The tool delivers substantive feedback in minutes instead of the hours typically required for manual review. This reduces code review turnaround time, allowing engineers to ship improvements faster.
