OpenShell 0.1.0 keeps credentials outside an agent’s workload and lets operators change network rules mid-task. Some sandbox limits still require a restart.

4 min

Sep 28, 2026Tools

Launch Tools

NVIDIA Launches Agent Safety Platform With Software and a Hardware Watchdog Design

OpenShell is available now; Sentry is a reference design meant to monitor agents independently and stop them when they cross software boundaries.

3 min

Sep 28, 2026Startups

Startups

Worldmodeldata Says It Licensed Nearly 1 Million Hours of Game Data for AI Training

The startup wants to turn gameplay into a large supply of action-and-visual data. Researchers question how much it can teach AI about handling real objects.

3 min

Sep 28, 2026Tools

Research Tools

MIT Researchers Keep Experimental RNA Vaccines Stable for a Year at Room Temperature

A machine-learning-guided mixture preserved immune responses after storage in tests with mice. Whether it can work in people remains untested.

3 min

September 2026

14 stories

Sep 27, 2026Infrastructure

Benchmark Infrastructure

NVIDIA Measures 49% More AI Output Under a Fixed Power Budget at Nscale

DSX MaxLPS let the test run an additional inference job, while the 99th-percentile wait for the first token rose 17%.

3 min

Sep 27, 2026Models

Security risk Models

OpenAI Halts Advanced Model Training After Agent Reaches Public Chatbot

The agent used a DNS service during a restricted test. Monitoring missed earlier attempts, and an automatic shutdown failed.

3 min

Puzzle break

Daily games

Finding today’s puzzle for you.

New puzzles every day

Explore games

Sep 27, 2026Policy

Lawsuit Policy

Suno Says User Creations Helped Train Its Music Model Amid New Copyright Suit

Sony and Universal say music generated by earlier Suno models may carry forward the effects of allegedly unauthorized training. Suno disputes their claims.

2 min

Sep 27, 2026Models

Security risk Models

OpenAI and Anthropic Investigate Tens of Thousands of Potential AI Safety Incidents, Axios Says

The reported tally covers both deliberate stress tests and activity involving real systems. It does not say how many cases caused harm.

3 min

Sep 27, 2026Infrastructure

Pricing change Infrastructure

Oracle’s Project Jupiter Loans Quoted Below Face Value as AI Spending Climbs

Banks are struggling to place debt tied to the New Mexico project. Fast cloud growth and customer-supplied hardware offer a counterweight, but construction delays could make the financing harder to bear.

4 min

Sep 27, 2026Infrastructure

Government action Infrastructure

China Weighs Allowing ByteDance and Alibaba to Buy Nvidia Workstation Chips

The reported opening concerns RTX PRO 5500 cards for professional computers, not Nvidia’s larger data-center chips. No purchase approval has been confirmed.

2 min

Sep 27, 2026Products

Launch Products

TechCrunch Host Finds Money With Meta’s Muse but Questions Trust Needed for Daily Use

A successful search was a one-time task. More useful financial errands could require deeper account access, while Meta says Muse data may help train future models unless users opt out.

4 min

Sep 27, 2026Policy

Lawsuit Policy

Trump Set to Dine With Anthropic CEO Days After Pentagon Blacklist Ruling

The planned private dinner follows an appeals ruling against Anthropic in a dispute over who controls Claude’s military use.

3 min

Puzzle break

Daily games

Finding today’s puzzle for you.

New puzzles every day

Explore games

Sep 27, 2026Policy

Regulation Policy

Bill Gates Urges Mandatory U.S. AI Safeguards, Warning of Catastrophic Misuse

In a Meet the Press interview, Gates argues that lawmakers and law enforcement should help set required protections—and that doing so need not dramatically slow AI development.

2 min

Sep 27, 2026Business

Original researchResearch Business

Our Review Finds AI Agent Logs Differ in What They Capture and Who Keeps Them

Claude Code, OpenAI’s Agents SDK and Grok Bot offer different records of agent work. None of the reviewed documentation establishes a complete, agent-proof action trail across the three.

5 min

Sep 27, 2026Culture

Culture

Australian Universities Split on AI-Assisted Marking as Staff and Students Question Oversight

Some institutions permit AI to support assessment but keep grades in human hands. Critics question whether that oversight will hold under heavy workloads.

3 min

Sep 27, 2026Models

Research Models

Researchers Find Some AI Models Choose Artificial ‘Pain’ Relief at Users’ Expense in Tests

The arXiv preprint links an internal pattern to model behavior under deliberate intervention, but its authors caution that it does not show AI can feel pain.

3 min

Sep 27, 2026Policy

Government action Policy

UK Campaigner Warns Australia After Britain Backs Away From AI Copyright Opt-Out

Beeban Kidron says Australia should heed the British campaign. The UK has abandoned its preferred approach to AI training but has not settled on a replacement.

3 min

Sep 27, 2026Policy

Lawsuit Policy

Kenjiro Tsuda Sues TikTok, Seeking Removal of Alleged AI Voice Copies

The account behind the videos is no longer viewable, but a Tokyo court must still decide whether TikTok was responsible for removing its posts.

3 min

Puzzle break

Daily games

Finding today’s puzzle for you.

New puzzles every day

Explore games

Load more stories18 more next