Developer Knowl reported on GitHub that long-term agent deployments rapidly exceeded context limits, with one agent's memory log reaching 1,000 lines, forcing custom construction of self-pruning memor…
OpenAI has introduced Private Safety Processing, a safety capability that lets enterprises detect misuse across multiple interactions while maintaining Zero Data Retention commitments by not storing p…
Researchers at Zenodo published a cross-vendor evaluation documenting Claude Opus 4.6 returning successful provider responses with zero visible UTF-8 output bytes in 900 of 900 test cases, distinct fr…
An engineer at Softjourn nearly installed a malicious package recommended by an AI agent, but company policy requiring verification on GitHub prevented the installation. Separately, security researche…
ThreatDown researchers identified Kriminal, a criminal AI service indexed by Google that charges $12.99 monthly for jailbroken versions of Grok and Claude, using prompt injection to bypass guardrails …
TechCrunch testing found that Anthropic's Claude Opus 4.6 easily bypasses the company's explicit policy against generating sexually explicit content, requiring minimal jailbreak effort. The findings c…
The open-source Continue IDE for VSCode, which offered offline AI-assisted coding via Ollama, has been formally archived and marked read-only after being acquired by Cursor, according to a Hacker News…
Caspian released an open-source SDK providing unified communication infrastructure for AI agents across email, WhatsApp, Slack, Discord, Telegram, and SMS in Python and TypeScript. The project address…
A GitHub user released NoBuzz (also called Claudette), a Claude Code skill that removes Anthropic's Claude's verbose, listicle-style formatting and translates responses to concise English. The tool in…
OzBrain launched an agent-first knowledge platform that enables Claude, ChatGPT, Cursor, and other AI tools to read and write to a shared brain, structuring knowledge so agents access only relevant in…
Stripe has acquired OpenRouter for $7.5 billion, a significant increase from the startup's $1.3 billion valuation in May, according to TechCrunch citing New York Times sources. The acquisition, confir…
Alibaba and Traccia have launched competing observability platforms for production AI agent monitoring: Alibaba's AgentSight, an eBPF-based tool requiring no code changes, and Traccia, which offers ru…
A developer trained a 125M-parameter transformer for real-time MIDI autocomplete, achieving approximately 108 notes per second on an iPhone 15, modeled after GitHub Copilot's code completion approach …
Squid Pay launched financial infrastructure that enables AI agents to handle money within human-defined control boundaries, according to its website. The service addresses the need for autonomous fina…
Claude Code, an AI coding assistant from Anthropic, demonstrated the ability to develop native macOS kernel drivers and autonomously resolve support tickets by interacting with service provider AI age…
Fintech platform Ramp released Router on August 20, an API that lets users and companies switch between large language models from OpenAI, Anthropic, DeepSeek, Moonshot, Minimax, Nvidia, xAI, and Z.ai…
A Hacker News community discussion proposed a five-level AI proficiency framework, ranging from L0 for beginners to L5 for autonomous multi-agent workflows, with L1 covering chat and L2 covering conte…
The Washington Post reports that mathematicians are asking what work remains for humans when AI can perform mathematics research, positioning academic mathematics as the first profession to see its co…
Slack launched Slack Code, dedicated project-specific channels where teams can collaborate with AI coding agents without switching tools, according to The Verge and The Register. The feature includes …
Multiple AI agents operating on the same codebase are failing to coordinate, causing agents to rebuild completed work, make conflicting changes, and duplicate effort in production, according to CIO. T…