Snippets.

// AI Briefing

October 5, 2026

AI Briefing

Washington wants a bigger hand on the AI wheel, and it picked a spy chief to hold it. Meanwhile Sam Altman is openly saying the world should accept some bad outcomes from AI, OpenAI is charging $500 a month for speed, and a Nvidia-backed startup is about to give American companies an open-weight answer to DeepSeek. Oh, and an AI just helped design a blood cancer drug candidate.

Trump Launches a 'Super Intelligence Force' Led by the Director of National Intelligence
01PolicyCBS News ↗

Trump Launches a 'Super Intelligence Force' Led by the Director of National Intelligence

President Trump announced a new federal AI task force on Sunday, naming Director of National Intelligence Jay Clayton to lead it alongside FTC Chair Andrew Ferguson, Pentagon CTO Emil Michael and OPM Director Scott Kupor. It reports directly to Trump and chief of staff Susie Wiles, and comes days after top AI lab leaders signed a voluntary self-policing accord at the White House. Putting an intelligence chief and the FTC chair in the same group signals AI is now being treated as a national security and consumer protection issue at once.

Altman Says the World Should 'Accept Some Bad Things' From AI, Breaking With Anthropic

Altman Says the World Should 'Accept Some Bad Things' From AI, Breaking With Anthropic

In an interview with Politico's Decoded newsletter, Sam Altman said there is 'a lot of daylight' between OpenAI and Anthropic on regulation. He rejected the idea that a single San Francisco lab should hold a dangerously powerful technology and decide how benefits are doled out, calling that trade-off unacceptable. It is the clearest public statement yet of the split over how much risk society should tolerate in exchange for broad access.

ChatGPT Gets a $500 Pro Tier While the $200 Plan Gets Its Usage Cut in Half

ChatGPT Gets a $500 Pro Tier While the $200 Plan Gets Its Usage Cut in Half

At DevDay, OpenAI launched a $500 per month Pro plan whose headline feature is an Ultrafast tier, up to 300 tokens per second in Codex, about 8x faster than standard. New $200 Pro subscribers from October 30 get half the Work and Codex allowance (10x Plus instead of 20x) and 100 weekly GPT-6 Pro messages instead of 200. Existing subscribers keep current limits through October 29 and get a one-time $2,500 usage credit. Speed is becoming the premium product.

Nvidia-Backed Reflection AI Is Set to Release Its First Open-Weight Model
04Open SourceRuntime Wire ↗

Nvidia-Backed Reflection AI Is Set to Release Its First Open-Weight Model

Axios reported that Reflection AI, led by ex-DeepMind researcher Misha Laskin, will soon release its first open-weight model, initially behind the best US systems. The pitch is an 'AI factory': models, software and compute that let governments and companies build AI on their own data and infrastructure, backed by Nvidia compute deals with SpaceX and Nebius. It is a bet that Western enterprises want a controllable open alternative to Chinese models like DeepSeek.

Insilico's AI Designs Potential Blood Cancer Drugs Using PROTAC Molecules

Insilico's AI Designs Potential Blood Cancer Drugs Using PROTAC Molecules

Insilico Medicine's Abu Dhabi team used AI to design a class of PROTAC compounds targeting blood cancers such as leukemias and lymphomas. The company typically takes 12 to 18 months to pick a candidate for further testing, and hopes AI lifts the odds in a field where nine in ten drugs reaching human trials fail. It is an early result, but a concrete example of AI compressing the earliest, slowest stage of drug discovery.

Test Your Understanding

Quiz

1 / 8

Who was named to lead the new 'Super Intelligence Force'?

Let's talk on WhatsApp