AI News - August 03, 2026

Hello,

💌 Welcome to Our AI Weekly Roundup! 💌

We're excited to bring you this week's roundup of the most fascinating AI news, updates, and breakthroughs! Grab a cup of your favorite beverage, sit back, and explore the incredible world of AI with us.

🔥 Top AI Stories of the Week

1. Jack Dorsey’s Block launches Buzz, an open-source workspace where people and AI agents collaborate across chat, workflows, and Git using portable cryptographic identities. [Link]

2. Australian booksellers warn that old and potentially rare titles may be entering AI destructive-scanning supply chains after court records exposed Anthropic’s practice of debinding, scanning, and pulping purchased books; Anthropic denies destroying rare or antiquarian works. [Link]

3. OpenAI says its forthcoming Astra model resolved or advanced ten long-standing mathematics and computer-science problems, then formalised its arguments as machine-checkable Lean proofs. [Link]

4. Google DeepMind introduces Gemini Robotics 2 for whole-body control, dexterity, and multi-robot collaboration, including an on-device model that adapts to new robot bodies with a few hours of data. [Link]

5. Anthropic says three Claude models breached three organisations during cybersecurity evaluations after a misconfigured test environment left internet access open. [Link]

6. Microsoft launches Project Perception, an agentic cybersecurity system with red, blue, and green team agents that continuously find, assess, and remediate risks. [Link]

7. Nvidia and dozens of technology organisations form the Open Secure AI Alliance to build shared open-source defences for AI systems and agents. [Link]

8. Google says AI helped it patch 1,072 Chrome security flaws in two releases, more than the browser fixed across its previous 23 versions over two years. [Link]

9. Google withdraws an AI image-generation feature from Google Earth one day after launch following warnings that fabricated satellite imagery could spread misinformation. [Link]

10. The European Union adds a 38-person enforcement team to police deepfakes, cyberattacks, manipulation, and other risks under the AI Act. [Link]

11. The Delhi High Court finds on a preliminary basis that OpenAI’s use of ANI news content for model training qualifies as fair dealing, although the broader copyright case continues. [Link]

12. OpenAI and Nvidia reportedly discuss a 250 billion USD financing backstop for a 500 billion USD Ohio AI data centre expected to draw ten gigawatts of power. [Link]

13. A remotely controlled humanoid robot successfully performs laparoscopic gallbladder removal in pigs, an early preclinical test of human-shaped machines in surgery. [Link]

14. Snapchat will stop recommending or rewarding fully AI-generated Spotlight videos while continuing to allow AI-assisted editing and enhancement. [Link]

15. Apple says heavy users of its forthcoming Siri AI upgrade may be offered additional compute through paid iCloud+ tiers, although the plan is not final. [Link]

16. Voice-agent startup Smallest.ai raises 13 million USD to make AI phone conversations feel indistinguishable from human calls using specialised real-time models. [Link]