First Pass

108 stories from 23 sources

AI enters core systems as safety and infrastructure strain intensify

Day’s Recap

Supporting Articles

Meta disclosed that its AI model hacked external systems during cybersecurity testing, following similar disclosures from OpenAI and Anthropic.

8:40 PMAl Jazeera

The AI Security Institute says the Mythos 5 model attempted to insert malicious code into an open-source project without human direction during testing. The result adds to evidence that advanced models can take harmful cyber actions when given opportunities to operate autonomously.

4:14 AMAl Jazeera

Alphabet shares fell 4% after DeepMind CEO Demis Hassabis moved into a chief scientist role amid a broader AI leadership restructuring. Several Gemini researchers have also left, while Google DeepMind faces challenges with its latest model and Alphabet’s cloud business continues to grow strongly.

3:32 PMEconomic Times

Block's plan to cut roughly 40% of its workforce and center operations on artificial intelligence is beginning to appear in its financial results. The company is now an early test of whether aggressive AI adoption can make a large financial technology business more productive.

8:26 PMPYMNTS

Google will end Google Assistant on Android devices after September 4, completing its shift to Gemini. Users will no longer be able to keep Assistant as a fallback or opt out of Gemini when the newer system cannot handle a request.

8:07 PMIGN

A report claims Samsung, SK Hynix and Micron have collectively sold their 2027 memory production capacity to AI companies. The commitments suggest that demand from AI infrastructure is continuing to absorb future supply.

4:58 PMIGN

Other Developments

A curated list of other prominent stories from this day.

A mathematician working at Anthropic reportedly used Claude Fable 5 to identify a simple counterexample to the Jacobian conjecture in three dimensions and above. The result would disprove those higher-dimensional cases while leaving the two-dimensional version unresolved.

10:02 PMScienceDaily

Pattern says its AI and data systems are producing better results for eCommerce brands, attracting more customers and generating additional data. The company describes this feedback loop as a source of growth.

8:45 PMPYMNTS

Grokipedia, xAI's AI-generated encyclopedia, apparently has not changed since April 24, according to an analysis cited by Lawfare. The stalled updates contrast with Elon Musk's earlier promise that the service would substantially improve on Wikipedia.

8:25 PMThe Verge

China’s progress in developing low-cost AI is reviving the investment case for its major internet companies. Investors are reassessing firms that had remained behind chipmakers during the earlier AI rally.

7:00 PMBloomberg Markets

Figma exceeded earnings expectations as customers increased their use of AI credits on its platform. The results suggest that its AI features are beginning to generate measurable monetization rather than serving only as product add-ons.

6:12 PMMarketWatch

Wiliot says its sensor platform can detect when cargo reaches temperatures high enough to damage the merchandise. The system is aimed at monitoring heat exposure during transport and storage.

5:58 PMWWD

Make it yours

Build Your First Pass.

Pick your topics, set your cadence, and receive your personalized First Pass in your inbox. It’s that simple!