The Pentagon is reportedly planning to adopt and weaponize frontier AI models with hacking capabilities, such as Anthropic's Claude Mythos Preview. This move comes despite Anthropic having been designated a "supply chain risk" by the Pentagon earlier this year due to a public disagreement and pending litigation.

The initiative was announced by Joshua Rudd, the leader of Cyber Command and director of the NSA. The task force aims to study how to safely deploy AI models across all missions and assess their use on "high-side" systems containing sensitive intelligence. Former Cyber Command deputy commander Charles Moore emphasized that AI tools are rapidly becoming essential for detecting threats, prioritizing vulnerabilities, accelerating decision-making, and conducting both defensive and offensive cyber operations.

Anthropic has previously indicated to Politico its openness to Mythos being used in offensive operations. The Trump Administration is also involved with Project Glasswing due to national-security risks posed by the model, and reportedly held veto power over Anthropic's attempts to expand the project to new companies and organizations.

👉 Read the full article on Gizmodo

The content of this TechCrunch article was inaccessible. Despite attempts to find alternative sources via simulated Google Search as per instructions, the tool's limitations prevented dynamic search and content retrieval. Therefore, a detailed summary based on the article's specific content cannot be provided.

Based on the title, it is highly probable that Sam Altman, a prominent figure in AI and former president of Y Combinator, made a significant offer to startups within the Y Combinator ecosystem. This offer likely pertains to resources, funding, or access to advanced AI models or technologies from companies like OpenAI, aiming to accelerate their development and integration of Generative AI.

👉 Read the full article on TechCrunch

AMD is aiming to preempt Nvidia's rumored laptop CPUs with its most powerful processor featuring the latest Ryzen AI Halo chips. The newly unveiled Ryzen AI Max+ Pro 495 chip boasts a Zen 5 CPU with 16 cores and 32 threads, and a boost clock speed of 5.2GHz.

While it utilizes the same RDNA 3.5 GPU microarchitecture as the previous generation, it supports up to 160GB of VRAM, pushing GPU memory further. AMD claims the Ryzen AI Max+ Pro 495 is its first x86 processor capable of running a 300 billion parameter AI model independently. Two other versions, the 12-core Ryzen AI Max Pro 490 and the 8-core Ryzen AI Max Pro 485, were also announced.

In a direct challenge to Nvidia's DGX Spark, AMD is launching a miniature "AI developer platform" called AMD Ryzen AI Halo, priced at $4,000. This small PC features the previous-gen Ryzen AI Max+ 395 chip, supports a 2TB SSD and 128GB of unified memory, and runs both Linux and Windows. AMD asserts its AI Max+ chip offers 4X generative AI workloads compared to an M4 Pro.

Preorders for the Ryzen AI Halo with the 395 chip are expected in June, with the 495+ version "coming soon." New devices from Asus, HP, and Lenovo featuring the 495+ chip are anticipated in Q3, setting up competition against Nvidia's upcoming laptop CPUs and Intel's Panther Lake chips.

👉 Read the full article on Gizmodo