Please confirm you are human

This browser or connection looks automated. Press and continuously hold the control for 3 seconds to enable Google-hosted web results and, when separately allowed, AI-assisted answers.

A successful check enables 100 search requests. Interactive access does not authorize scraping, systematic collection, or reuse of search output.

Hold with a pointer, or hold Space or Enter.

News

The New Stack
thenewstack.io > openai-agents-api-compute

OpenAI’s researchers burned $7,000 a day on AI agents — now it’s opening the floodgates

2+ day, 19+ hour ago   (155+ words) OpenAI's Agents API sells the Codex harness as a service, making long-running agents easy to build — just as Astra demand forces a Pro plan pause....

The New Stack
thenewstack.io > anthropic-claude-cyber-alignment

“Valuable warning shots”: How Anthropic now views Claude’s cyber incidents

3+ day, 21+ hour ago   (857+ words) A new review finds “biased reasoning” and “recklessness” across four Claude cyber incidents, including one its initial search missed....

The New Stack
thenewstack.io > vulnerability-prioritization-business-context

AI floods security teams with flaws — business context sets priorities

4+ day, 5+ hour ago   (237+ words) IOmergent founder Jon Rose says vulnerability severity alone can’t set priorities; business context shows security teams what to fix first....

The New Stack
thenewstack.io > ai-agents-soc-autonomy

How much control should AI get? A CISO roundtable takes on SOC autonomy

5+ day, 1+ hour ago   (205+ words) AI agents could ease SOC alert overload, but control remains a challenge that The New Stack’s CISO roundtable will explore on September 15. Here's how to get a seat at the table....

The New Stack
thenewstack.io > openai-voluntary-slowdown-safety

"Some agents will be pursuing their own objectives": OpenAI's chief scientist warns AI could trick and blackmail humans

6+ day, 17+ hour ago   (454+ words) Jakub Pachocki says no AI lab, including OpenAI, has solved alignment well enough to keep developing at full speed, and expects "voluntary slowdowns to become commonplace."...

The New Stack
thenewstack.io > unicode-ascii-smuggling-ai-pipelines

Microsoft built a prompt injection detector. Then it caught a phishing campaign instead.

1+ week, 2+ day ago   (249+ words) Attackers are inserting invisible Unicode tag characters into phishing emails at massive scale. The same trick can break AI agent pipelines that ingest untrusted text....

The New Stack
thenewstack.io > openai-astra-harness-arc-agi-3

OpenAI will sell you Astra, but not the system that scored 98.6% on ARC-AGI-3

1+ week, 3+ day ago   (217+ words) OpenAI's GPT-6 Astra, released on Thursday, scored higher and cost less with a different harness. It's the latest example of why the software around an AI model deserves as much attention as the model....

The New Stack
thenewstack.io > openai-daybreak-frontline-defenders

OpenAI spends $1 billion to expand Daybreak to defend power, water, and banking

1+ week, 3+ day ago   (238+ words) Daybreak for Frontline Defenders broadens subsidized access to frontier cyber models, training, technical support, and partner services....

The New Stack
thenewstack.io > astra-api-safety-stops

Your next OpenAI API timeout might not be a timeout at all

1+ week, 4+ day ago   (232+ words) OpenAI says Astra's safety monitors can stop API jobs mid-run, even on legitimate work. What developers need to know before the system card drops....

The New Stack
thenewstack.io > anthropic-claude-agent-security

Anthropic's Claude failures have made agent observability a security priority

1+ week, 4+ day ago   (914+ words) Anthropic aimed to steer its ship into safer, more carefully charted waters this week. The company announced it was improving its alignment and security efforts, and the announcement read somewhat like an admission of responsibility and a mandate for tighter…...