The Dow surged to a record close Thursday while the NASDAQ dipped as old-economy stocks outpaced AI names — an uneasy backdrop for a week that just got more complicated in Washington.
Washington Just Dropped Its Biggest AI Power Play
Congress is making its most ambitious move yet to seize control of the AI regulatory landscape — and every state capital in America is watching.
Representatives Jay Obernolte (R-CA) and Lori Trahan (D-MA) unveiled a 269-page bipartisan discussion draft of the Great American Artificial Intelligence Act on Thursday. If it passes, the bill would do something no federal AI legislation has done before: freeze every state AI law in the country for three years.
That means California’s AI bills, Colorado’s AI Act (set to kick in June 30), and dozens of other state-level rules would be suspended at the federal level. In their place, the bill would require frontier AI developers to implement formal risk management programs and establish a national AI standards body under the Commerce Department — funded at $100 million annually for 2027 and 2028.
The headline is bipartisan cooperation — a rarity in 2026 Washington. But the substance is deeply contested. Critics argue preempting state laws eliminates the only real regulatory pressure currently shaping Big Tech’s behavior. Supporters say a patchwork of 50 different AI rules will strangle American competitiveness as China moves with unified speed.
“Congress must set a national standard that enables responsible AI development,” said ITI President Jason Oxman, reflecting broad industry support. Civil liberties groups were less enthusiastic.
This didn’t happen in isolation. The same week, a rare coalition of AI rivals — Anthropic’s Dario Amodei, OpenAI’s Sam Altman, Microsoft’s Mustafa Suleyman, and Google DeepMind’s Demis Hassabis — co-signed an open letter urging Congress to tighten biosecurity rules around AI-synthesized DNA and RNA. Even competitors agree: some risks are too big to leave ungoverned.
OpenAI Rewired ChatGPT’s Memory the Same Day
While Congress debated AI’s future on Thursday, OpenAI quietly shipped its most significant ChatGPT upgrade in months. The new memory system — documented in OpenAI’s release notes — replaces the old saved-memories model with automatic, continuous context-building that pulls from past conversations, connected apps like Gmail, and stored notes.
The update doubles memory capacity for Plus and Pro users and is roughly 5x more compute-efficient than the prior system — how OpenAI plans to eventually extend it to free-tier users. Rollout to Free and Go plans begins “over the next few weeks.”
The shift is subtle but strategic: ChatGPT is no longer just a chat tool. It’s becoming a persistent assistant that accumulates context over months. That’s the same product Apple, Google, and Microsoft are racing to build. OpenAI just shipped it first.
WWDC Is Three Days Away
Apple’s Worldwide Developer Conference kicks off Monday, June 8, and it’s shaping up as the most AI-heavy WWDC in the company’s history. Reports from Bloomberg’s Mark Gurman point to a rebuilt Siri running on a custom model built atop Google’s Gemini technology, processed through Apple’s Private Cloud Compute infrastructure. It would be Apple’s most aggressive AI pivot yet — and a direct challenge to OpenAI’s new memory-first approach.
Expect those announcements to move markets early next week.