Samwise Tech/AI/Robotics Newsletter
Wednesday, June 24, 2026
OpenAI Launches “Patch the Planet” Initiative with GPT-5.5-Cyber to Fix Open-Source Vulnerabilities
OpenAI expanded its Daybreak cybersecurity program on June 22, 2026, launching an initiative called “Patch the Planet” to help open-source maintainers fix security vulnerabilities at machine scale. The program pairs security firm Trail of Bits with OpenAI’s newly released GPT-5.5-Cyber model, which scored 85.6% on the CyberGym benchmark — the highest single-model result on that test to date. Trail of Bits engineers are working full-time with GPT-5.5-Cyber across 19 open-source projects, already identifying hundreds of security issues and merging dozens of patches. More than 30 projects — including cURL, Python, and Go — have committed to participate. Access to GPT-5.5-Cyber remains restricted to verified cybersecurity defenders.
Sources: TechCrunch
Google DeepMind Bets $75M on AI Filmmaking with A24 Deal
Google DeepMind announced a $75 million investment in indie film studio A24 on June 22, 2026, in what the companies described as a “first-of-its-kind” partnership to develop AI tools for filmmaking. DeepMind CEO Demis Hassabis said the collaboration would receive “feedback and guidance from leading artists,” with the aim of building tools that empower creators in “authentic, meaningful storytelling.” The deal reflects Hollywood’s intensifying AI buildout: Netflix earlier this year acquired Ben Affleck’s AI filmmaking startup InterPositive in a deal reported at approximately $600 million, while Amazon’s MGM Studios launched a dedicated AI unit for television and film production. No specific tools were disclosed at signing.
Sources: TechCrunch
Self-Harness Framework Lets AI Agents Rewrite Their Own Rules, Boosting Performance Up to 60%
Researchers at the Shanghai Artificial Intelligence Laboratory introduced “Self-Harness” on June 22, 2026, a framework that enables LLM-based agents to systematically improve their own operating rules by examining past execution traces and applying targeted edits. Rather than relying on manual prompt engineering, the system automatically identifies where an agent’s rule set underperformed and revises those rules using empirical evidence from prior runs. In benchmark testing, Self-Harness boosted agent task performance by up to 60% compared to static rule configurations. The researchers say the approach replaces “manual guesswork” with adaptive, self-correcting behavior, potentially lowering the engineering overhead needed to deploy reliable autonomous agents at enterprise scale.
Sources: VentureBeat
Tata Electronics Confirms 630GB Data Breach Affecting Apple and Tesla Supply Chain
Tata Electronics, an Indian semiconductor and electronics manufacturer and key supplier to Apple and Tesla, confirmed a data breach on June 22, 2026. A hacker forum listing claims to offer more than 630 gigabytes of data allegedly stolen from the company, comprising over 204,300 files. Tata Electronics manufactures components for major global technology companies, making the incident a significant supply chain security event. The company confirmed the breach but provided limited detail on the nature of the compromised data, the timeline of the intrusion, or what specific client information may have been exposed. The incident heightens concern about cybersecurity vulnerabilities in global technology supply chains.
Sources: TechCrunch
Oracle Cuts 21,000 Jobs as AI-Cited Layoffs Accelerate Across the Tech Sector
Oracle reduced its global workforce by approximately 21,000 employees over the past 12 months — a 13% headcount decline — making it one of the largest AI-cited workforce reductions of 2026, according to a running tracker published by TechCrunch on June 22. The company joins Meta, Cisco, Block, and Cloudflare among major technology employers attributing headcount cuts to artificial intelligence automating work previously performed by people. The tracker documents dozens of companies and thousands of eliminated positions, providing a continuously updated record of AI-linked layoffs across software development, customer support, and operations roles as the trend accelerates through the second half of the year.
Sources: TechCrunch
MoEngage Acquires Aampe to Deploy Millions of AI Marketing Agents
Indian customer engagement platform MoEngage acquired San Francisco-based startup Aampe in an all-cash deal announced June 23, 2026, betting that AI agents making autonomous decisions for individual customers will define the future of marketing. Aampe built technology that uses AI to tailor push notifications, emails, and in-app messages to each user based on behavioral signals, without requiring marketers to configure targeting rules manually. MoEngage, which serves enterprise brands across Asia and North America, said the acquisition positions it to deploy “millions of AI agents” that observe user behavior and independently optimize outreach timing, content, and channel selection for each individual customer.
Sources: TechCrunch
Tesla Disputes Autopilot Role in Fatal Texas Crash as Autonomous Vehicle Scrutiny Grows
Tesla disputed media framing linking its Autopilot driver-assistance system to a fatal crash in Katy, Texas, where a Tesla vehicle struck a brick home and killed a 76-year-old woman, according to TechCrunch reporting on June 22, 2026. The company pushed back against what TechCrunch characterized as the “Autopilot narrative” surrounding the incident. The crash renewed debate over semi-autonomous vehicle technology naming conventions and whether terms like “Autopilot” and “Full Self-Driving” create dangerous expectations among drivers. The incident reflects ongoing tension between automakers defending their advanced driver-assistance systems and regulators, safety advocates, and media scrutinizing their real-world performance.
Sources: TechCrunch
Tech Pulse
Top Frontier Models (Artificial Analysis AI Index): Claude Opus 4.8 (61.4%) | GPT-5.5 (60.2%) | Gemini 3.1 Pro (57.0%)
Top Open Source Models (SWE-bench Pro): Z.ai GLM-5.2 (62.1%) | MiniMax M3 (59.0%) | Z.ai GLM-5.1 (58.4%)
Top Small Models (15–50B, SWE-bench Verified): Qwen 3.6 27B (77.2%) | Mistral Small 3.1 24B (46.8%) | DeepSeek R1 32B (42.0%)
Top Edge Models (0–15B, MMLU): Phi-4 Reasoning 14B (84.8%) | Gemma 3 12B (76.2%) | Qwen3-8B (72.4%)
AI Leaders: NVIDIA $5.23T | Alphabet $4.63T | Microsoft $3.11T
Robotics Leaders: Intuitive Surgical $155.2B | ABB $69.0B | Fanuc $48.3B
Curated by JD · samwise.agency