AI Daily Digest 2026-08-09: Claude Code Multi-Terminal Sync, OpenAI Acquires NextSlide

🤖 Claude Code Introduces Multi-Terminal Session Communication and Sets Auto Mode as Default

Anthropic continues to cement its leadership in the AI agent developer tools space with two major updates for Claude Code. First is inter-session communication on macOS and Linux. Parallel instances of Claude Code running across different terminal windows can now exchange messages, share working context, pass task coordinates, and inspect each other’s execution status in real time. This capability allows developers to orchestrate a coordinated fleet of AI coding agents that break down complex repositories into parallel sub-tasks without losing context across process boundaries.

Additionally, Anthropic announced that starting August 14, Auto Mode will become the default setting for Pro, Max, and Team plans. The decision comes after extensive field testing showing that Claude Code’s enhanced classifier catches and intercepts up to 89% of hazardous CLI commands (such as inadvertent root directory deletions or unauthorized credential exposure). Making Auto Mode the default protects developers from accidental approval mistakes while keeping agent execution fast and highly autonomous.

Source: The Decoder

💼 OpenAI Acquires Presentation Startup NextSlide to Power ChatGPT Slides

OpenAI has acquired NextSlide, a startup known for AI-powered presentation creation tools. The entire NextSlide engineering and design team is joining OpenAI to build native presentation and deck creation capabilities directly into the ChatGPT ecosystem. The move is widely seen as a direct strategic response to Microsoft Copilot in PowerPoint and Google Slides AI for enterprise office workflows.

This acquisition signals ChatGPT’s rapid evolution from a text and code conversational interface into a complete multimodal productivity engine. With NextSlide’s core technology integrated, users will soon be able to ask ChatGPT to instantly transform research notes, data analysis, or strategic outlines into professionally formatted presentation decks complete with visual layouts, structured slides, and clear data graphics within seconds.

Source: TechCrunch

🛡️ Fields Medalist Who Warned of AI Extinction Joins OpenAI Safety Team

Jacob Tsimerman, a distinguished mathematician from the University of Toronto who recently won the prestigious Fields Medal, is leaving academia to join OpenAI’s AI Safety alignment team. The announcement has drawn widespread interest across the scientific community because Tsimerman recently co-authored a prominent mathematical paper analyzing catastrophic existential risks where superintelligent AI systems could lead to human extinction.

At OpenAI, Tsimerman will bring advanced mathematical tools and algebraic geometry to bear on quantitative safety modeling for next-generation frontier models. The transition of a world-class theoretical mathematician from voicing existential concerns to working directly inside a leading AI lab reflects an industry-wide push to recruit top-tier theoretical minds to solve complex alignment challenges as models approach Artificial General Intelligence (AGI).

Source: The Decoder

📚 Readers Rate AI Short Stories Higher Than Human Ones — Until Told a Machine Wrote Them

A new study involving over 2,500 human participants reveals a surprising psychological phenomenon: readers rated ChatGPT-generated short stories higher in literary quality, narrative engagement, and emotional depth than human-authored stories during double-blind evaluations. In blind testing, participants could not reliably distinguish human writing from machine-generated text, performing no better than random guessing.

However, the ratings flipped dramatically as soon as participants were informed that a story was generated by AI. Evaluation scores dropped significantly due to subconscious anti-AI bias among readers. The findings demonstrate that while modern large language models have reached or surpassed human benchmarks in technical storytelling and prose generation, public perception and psychological bias remain the primary hurdles for AI creative adoption.

Source: The Decoder

📐 Backflip AI Converts 3D Scans into Editable Parametric CAD Models in Minutes

Startup Backflip AI has launched a groundbreaking neural model capable of converting raw, unstructured 3D mesh scans directly into fully editable, parametric CAD geometries (STEP/IGES formats). Historically, reverse-engineering a physical object from a 3D scan required mechanical engineers hours of painstaking manual reconstruction in software like SolidWorks or Fusion 360.

Backflip AI reduces that workflow to under 3 minutes with high geometric precision, automatically reconstructing parametric surfaces, technical holes, and smooth curvature fillets. Industry observers view the release as a game-changer for hardware manufacturing, rapid 3D printing prototyping, and industrial digital twin automation, speeding up physical design iteration by orders of magnitude.

Source: The Decoder