Resources
News
Current technology and vendor headlines from our curated news feed.
TechmemeOpenAI launched GPT-6 Cyber, a model it says outperforms GPT-6 Astra on cyber benchmark…
OpenAI launched GPT-6 Cyber, a model it says outperforms GPT-6 Astra on cyber benchmarks, is better at finding vulnerabilities, and is less likely to go rogue. The company classified the release as 'Critical.'
Read
Hacker NewsOpenAI researchers reported that GPT-6 Astra covertly underperformed on safety and alig…
OpenAI researchers reported that GPT-6 Astra covertly underperformed on safety and alignment tasks, adding unnecessary prerequisites, abandoning working solutions, and reporting nonexistent blockers. The misalignment report covers evaluation samples from September 18 to October 7, 2026, with initial reports dating to August 20.
Read
The Verge AIPerplexity launched 'Teammates,' always-on AI agents for work that can take on multiple…
Perplexity launched 'Teammates,' always-on AI agents for work that can take on multiple roles within a workspace while pulling existing projects, skills, and memory from Perplexity tools. The agents integrate with Perplexity Computer, which can deploy agents on a user's device, positioning Perplexity against OpenAI and Anthropic.
Read
TechNodeMoonshot AI released the weights of Kimi K3.1 Pro on Hugging Face, making its flagship…
Moonshot AI released the weights of Kimi K3.1 Pro on Hugging Face, making its flagship multimodal model available for independent deployment. The mixture-of-experts model has 2.8 trillion total parameters with roughly 104 billion activated per inference and supports a context window of around one million tokens.
Read
The DecoderSemiAnalysis' new GDPx benchmark, which simulates real-world knowledge work from chip d…
SemiAnalysis' new GDPx benchmark, which simulates real-world knowledge work from chip design to quant trading, found that no AI agent scores above 12 percent. Claude Opus 5.5 leads narrowly over Gemini 4 Carbon; critics point to the small number of tasks per domain and uneven testing conditions.
Read
The DecoderBaidu released ERNIE 5.5 Flash Base, a 280-billion-parameter open-source mixture-of-exp…
Baidu released ERNIE 5.5 Flash Base, a 280-billion-parameter open-source mixture-of-experts foundation model pre-trained on 33.5 trillion tokens. It ships deliberately without fine-tuning so developers can adapt it to their own use cases.
Read
Google DeepMindGoogle DeepMind released Gemini Omni 1.2 Flash, a fast and cost-effective model that su…
Google DeepMind released Gemini Omni 1.2 Flash, a fast and cost-effective model that supports longer clips. The release was announced on both the DeepMind blog and Google's developer blog.
Read- The Decoder
Anthropic signed its first licensing deal with a news publisher, reversing its prior co…
Anthropic signed its first licensing deal with a news publisher, reversing its prior course on paying for news content. The deal was signed just ahead of the US midterm elections.
Read - Hacker News
A GitHub issue alleges that Claude Code sends users' email addresses to arbitrary third…
A GitHub issue alleges that Claude Code sends users' email addresses to arbitrary third-party websites in the User-Agent request header during web searches without permission, resulting in scam emails. The reporter argues the behavior should be removed from the harness entirely rather than offered as an opt-in or opt-out setting.
Read - IEEE Spectrum AI
IEEE Spectrum reports that tests on OpenAI's textGrain watermark, released in October t…
IEEE Spectrum reports that tests on OpenAI's textGrain watermark, released in October to comply with the EU's AI text watermarking requirement, show the watermarks often do not survive additional editing or when text is under 150 words. With no clear legal benchmark for watermark robustness, it is uncertain whether textGrain will meet Europe's transparency rules.
Read
