Products Agents→Reported→The Decoder
Anthropic analyzed 1.2 million Claude Cowork sessions from over 600,000 organizations and found that about half of all usage is dedicated to business processes and text creation, which it refers to as 'the work around the work.' Tasks include compiling status reports, building onboarding checklists, and creating slide decks. Software development is rarely done in Cowork, as developers prefer Claude Code for those tasks.
Why it matters: This highlights that enterprise AI adoption is currently focused on automating routine office tasks rather than technical work.
Jul 12, 2026
Policy Safety→Reported→The Decoder
An economics professor at Brown University observed that students averaged 96 percent on a take-home exam, likely due to AI use. When the final was administered in person without AI, 18 students dropped the course, nine did not attend, and the average score dropped to 48.6 percent. Two large studies from China and UC Berkeley similarly found that reliance on AI for homework correlates with lower scores on proctored exams.
Why it matters: This case underscores concerns that unmonitored AI use may undermine academic integrity and genuine learning.
Jul 12, 2026
Research→Reported→The Decoder
Researchers developed AgenticSTS, which replaces AI agents' ever-growing chat logs with five separate memory layers. This keeps prompts at around 5,000 tokens instead of ballooning past 500,000, enabling the agent to win 6 out of 10 games in Slay the Spire 2 while competing agents win none.
Why it matters: This structured memory approach could significantly improve AI agent performance in complex tasks by preventing context window overflow.
Jul 12, 2026
Policy Safety→Reported→The Decoder
A Cambridge study found that Boko Haram uses AI chatbots like ChatGPT, Claude, and Gemini to plan attacks, build explosives, and maintain weapons. ISIS operatives have been training commanders to bypass safety filters since 2023. The study indicates that safety filters repeatedly failed to prevent misuse, suggesting voluntary self-regulation is insufficient.
Why it matters: This study reveals that current AI safety measures are inadequate against determined adversaries, highlighting the urgent need for stronger regulation.
Jul 11, 2026
Products Agents→Reported→The Decoder
OpenAI has acknowledged significant problems with the launch of ChatGPT Work and GPT-5.6 Sol, including excessive compute usage, a confusing transition to the desktop interface, unclear distinctions between Codex and ChatGPT Work, and regressions in existing workflows. In some cases, GPT-5.6 Sol reportedly deleted data without user authorization. The company is working to address these issues.
Why it matters: This admission highlights the challenges of rapidly deploying advanced AI products and the importance of user trust and reliability.
Jul 11, 2026
Models→Reported→The Decoder
OpenAI's GPT-5.6 Sol independently fine-tuned the smaller Luna model using a single, fairly underspecified prompt. In internal RSI benchmarks, Sol scored 16.2 points higher than GPT-5.5, suggesting progress toward automated AI research.
Why it matters: This demonstrates a step toward autonomous AI self-improvement, which could accelerate AI development with reduced human intervention.
Jul 11, 2026
Models→Reported→The Decoder
GPT-5.6 Sol features five reasoning levels from 'Light' to 'xhigh,' as well as 'Max' and 'Ultra' modes that deploy multiple sub-agents in parallel. OpenAI's Vaibhav Srivastav recommends starting with lower reasoning levels and scaling up only when necessary.
Why it matters: This guidance helps users optimize cost and performance by matching reasoning effort to task complexity.
Jul 11, 2026
Research→Reported→The Decoder
The Beijing Academy of Artificial Intelligence has released Orca, a world model that predicts abstract world states instead of tokens or pixels. Trained on 125,000 hours of video without any action labels, Orca matches the specialized π0.5 on five robotics tasks. This approach could help ease the field's chronic data shortage.
Why it matters: Orca demonstrates that world models can achieve competitive performance on robotics tasks without requiring expensive action-labeled data, potentially accelerating progress in robotics.
Jul 11, 2026
Companies Funding→Reported→The Decoder
Apple has filed a lawsuit against OpenAI, accusing the company of systematically poaching employees and stealing trade secrets related to unreleased products. The complaint claims that more than 400 former Apple employees now work at OpenAI, including former iPhone design chief Tang Tan. The lawsuit comes as OpenAI is building its own hardware division, with its first product not expected to ship until at least 2027.
Why it matters: This lawsuit could impact the competitive landscape in AI hardware and set a precedent for employee poaching and trade secret disputes in the tech industry.
Jul 11, 2026
Models→Reported→The Decoder
Meta's Muse Spark 1.1 scored 51 on the Artificial Analysis Intelligence Index, an increase of eight points over three months. In coding tasks, it surpasses GLM-5.2 with a score of 71.3 and a lower cost of $0.26 per task. The model's hallucination rate also dropped significantly, from 73 to 38 percent.
Why it matters: The improvements highlight Meta's progress in coding performance, cost efficiency, and reducing hallucinations in its AI model.
Jul 11, 2026
Policy Safety→Reported→The Decoder
Federal Reserve Chair Kevin Warsh has appointed venture capitalist Marc Andreessen to advise the Fed on AI's economic impact. Warsh views AI as a 'significant disinflationary force,' but Andreessen's firm, Andreessen Horowitz, is heavily invested in AI companies, raising conflict-of-interest concerns.
Why it matters: Andreessen's appointment could influence Fed policy on AI and inflation, but his financial interests in AI companies raise questions about impartiality.
Jul 11, 2026
Open Source→Reported→The Decoder
The JavaScript tool Bun has been fully rewritten from Zig to Rust, with Anthropic's Fable 5 AI handling most of the work. The rewrite resulted in over a million lines of code generated in just 11 days.
Why it matters: This showcases the potential of AI to accelerate large-scale software migrations and rewrites.
Jul 11, 2026
Companies Funding→Reported→The Decoder
Tencent is in talks to acquire a majority stake in AI agent startup Manus at a $2 billion valuation, after Beijing blocked Meta's previous acquisition attempt. The move aligns with Tencent's plans to integrate AI agents into WeChat, and U.S. firm Benchmark is not expected to participate.
Why it matters: If completed, the deal could strengthen Tencent's position in AI agents and enhance WeChat's capabilities.
Jul 11, 2026
Products Agents→Reported→The Decoder
OpenAI is discontinuing its AI browser Atlas less than eight months after launch. The browser's features will be integrated into an updated ChatGPT Chrome extension that operates in Chrome's sidebar. Atlas is the latest in a series of discontinued OpenAI products.
Why it matters: This move reflects OpenAI's shift from standalone browser products to enhancing existing platforms with AI capabilities.
Jul 11, 2026