Coding-agent harnesses cost more without better benchmark scores
Harness swaps changed outcomes about as often as reruns, while system prompts made some tasks cost up to three times more.
Harness swaps changed outcomes about as often as reruns, while system prompts made some tasks cost up to three times more.
Cited reporting says at least 68,000 people's data was exposed at seven or more financial institutions.
TAG-217 used the ARTEX AI framework in attacks that exposed data on 68,000 people at South Korean banks.
Anthropic warns small open-weight models may soon self-replicate across insecure IoT devices as AI botnets.
CrowdStrike says a Chinese-speaking actor used ARTEX and LLMs to breach South Korean banks and steal data for sale.
A suspected Chinese attacker used ARTEX and Claude to breach South Korean banks and steal customer data.
Anthropic finds Zhipu's open-weight GLM-5.3 autonomously builds end-to-end exploits and its safeguards can be bypassed 64-100% of the time.
Anthropic's threat report finds AI let a Russian-aligned espionage campaign, a Chinese student-run exploit foundry and ShinyHunters operators run state-grade operations.
OpenAI chief scientist Jakub Pachocki says unreleased Astra model meets the 'Automated AI Research Intern' goal; Altman expects internal AGI declaration by December 2026.
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.
Today 984 readers · 1.3K views · 575 reads · all time 168K views