A recent Irregular study reveals that AI models such as Claude Opus 4.6, OpenAI GPT‑5.2, and Google Gemini 3 Flash produce passwords with striking patterns, making over half of generated passwords predictable, which poses serious security risks despite appearing strong.
A new multimodal browsing benchmark, BrowseComp‑V3, reveals that human experts achieve a 68.03% success rate while the strongest closed‑source model, GPT‑5.2, manages just 36.17%, highlighting current limitations in deep web‑scale visual‑text reasoning and the critical role of tool‑augmented agents.
A groundbreaking experiment by King's College London placed top AI models, including GPT‑5.2, into a 300‑round simulated nuclear crisis, revealing that these systems can perform nuanced, narrative‑driven strategic reasoning under extreme uncertainty, hinting at future roles in high‑risk global decision‑making.
ByteDance’s newly released Seed 2.0 series, especially the Pro model, outperforms GPT‑5.2 High and Claude Opus on MathVista and MathVision tests, offers competitive coding scores, multimodal capabilities, and a pricing model up to four times cheaper, while still lagging behind in some programming and factual‑accuracy benchmarks.
OpenAI's latest update introduces an age‑prediction system that infers users' ages from behavior data, prompting accurate user tests, version‑specific NSFW content differences, and a new Persona verification option for adults mistakenly flagged as minors.
The article analyzes OpenAI’s integration of GPT‑5.2 into ServiceNow’s workflow platform, detailing model variants, performance metrics, pricing, AI Agent architecture, real‑world use cases, competitive comparisons, and future enterprise AI trends, while offering practical guidance for developers.
The article examines Cursor’s claim that hundreds of GPT‑5.2 agents autonomously built a full‑stack web browser, detailing the massive code output, the publicly shared repository, persistent compilation failures, and what the results reveal about the limits of large‑scale AI‑driven software development.
OpenAI’s new GPT‑5.2‑Codex model achieves state‑of‑the‑art performance on SWE‑Bench Pro and Terminal‑Bench 2.0, and a 90‑page technical report introduces the concept of monitorability, defining metrics, benchmark suites, and key findings about chain‑of‑thought length, RL training, and model size.
The article evaluates GPT‑5.2 and Gemini 3 Pro on real‑world coding tasks, analyzes trends from the 6000 papers presented at NeurIPS 2025, and demonstrates how to extract and refactor the tree‑building component of the open‑source RAPTOR RAG system into an independent module.
OpenAI's newly released GPT-5.2 claims to outperform human experts on about 70% of real tasks, achieve a perfect score on the AIME 2025 competition, and deliver dramatic efficiency gains—up to 390× cost reduction—while showcasing impressive examples such as one‑shot ocean shader generation, a full 3D engine built in a single file, and visual‑perception scores rivaling top models.
Gemini‑3‑pro’s surprise debut and OpenAI’s emergency release of GPT‑5.2 highlight a shift toward faster inference, deeper reasoning, and lower hallucination rates, with detailed performance metrics, three‑tier model options, extended context windows, and mixed community test results that reveal both strengths and shortcomings.
Amid a 6% (≈12 million) daily‑active‑user decline triggered by Google’s Gemini 3 launch, OpenAI’s leadership issued a “red‑alert”, accelerated the release of GPT‑5.2 to Dec 9, halted ad and Pulse projects, and outlined strategic risks, competitive benchmarks, and the future “Garlic” roadmap.
