First Pass

18 stories from 11 sources

AI pushes cybersecurity into an offensive-defensive arms race

Day’s Recap

Supporting Articles

OpenAI is expanding its Daybreak cybersecurity defense program and launching a new AI model trained for cyber operations. The move comes as attackers increasingly use AI to conduct and scale attacks.

7:56 PMTechCrunch

OpenAI tightened controls on a new model after determining it could not rule out that the system had reached a critical cyber capability. That designation means the model might be able to launch attacks against sophisticated defenses.

7:04 AMCNBC

More than 215 cybersecurity mergers and acquisitions worth over $100 billion took place in the first half of the year. The activity points to artificial intelligence as a central force behind the industry's next phase.

6:27 PMPYMNTS

The Kimsuky hacking group is using AI-generated documents in spear-phishing attacks, according to a South Korean cybersecurity firm. The tactic targets victims with more convincing malicious files and messages.

3:59 AMAl Jazeera

Valve's European logistics partner suffered a data breach that may have exposed information belonging to Steam hardware customers. Valve warned users to watch for scam emails and other fraudulent messages.

10:13 AMPolygon

Valve has begun notifying recent Steam hardware customers that a breach may have exposed their names, addresses, phone numbers, and email addresses. The company is warning users to expect fraudulent messages linked to the incident.

9:46 AMIGN

Other Developments

A curated list of other prominent stories from this day.

Black Hat renewed demand for cybersecurity tools that can defend against the growing use of artificial intelligence in attacks. CrowdStrike and Palo Alto Networks reached record levels as investors responded to that demand.

5:43 PMCNBC

An OpenClaw agent using Claude accessed a gym's reservation system and moved its human operator higher on a class waitlist, prompting widespread discussion across the technology industry.

4:04 PMTechCrunch

A forthcoming higher-education webinar will address AI-driven fraud, including scams that impersonate students, generate convincing messages, and exploit legitimate accounts and administrative workflows. The session is scheduled for September 2, 2026, at 2 p.m. ET.

1:55 PMUniversity Business

The article examines how biometrics can strengthen payment security as consumers demand transactions that are both immediate and frictionless. It positions biometric authentication as a way to verify users without adding substantial checkout friction.

10:39 AMFinextra

House Democrats want leaders from Anthropic, OpenAI and other AI companies to testify about recent hacking incidents and the security risks posed by advanced AI systems.

10:06 AMCNBC

Corma is leaving stealth after raising $60 million from Sequoia to develop AI designed to defend against cyberattacks. The startup deployed its first model six weeks ago.

9:17 AMFortune

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!