Dave Eggers showed up at OpenAI’s headquarters with no intention to flatter. The author had slammed an AI-generated short story months earlier as “pastiche nonsense” and doubted an AI novel could be readable. OpenAI still invited him to speak to about 200 employees. Eggers didn’t hold back.
He told the staff, “The effect of ChatGPT on educators’ lives is catastrophic.” According to Eggers, ChatGPT has made teachers’ work infinitely harder than two years ago. When a staffer admitted to using ChatGPT for poetry, Eggers bluntly advised, “Why don’t you leave your house, leave the screen, meet some people, get yourself in a writing group?” His parting shot? “I had a ball.”
Meanwhile, SpaceX’s latest Starship launch failed again. Elon Musk confirmed, “Some of the engines didn’t start, triggering an automatic launch abort.” The company is offloading propellant and aims to try again in a few days. The abort hit Wall Street hard—SpaceX shares slid more than 4% in after-hours trading, dropping below their $135 IPO price. SpaceX’s June 12 public debut was historic, raising over $85 billion and putting the company alongside Amazon and Microsoft in valuation.
On the cybersecurity front, OpenAI quietly unveiled GPT-Red, a new AI super-hacker built to test defenses against cyberattacks. Trained through self-play, GPT-Red repeatedly attacked and defended models, evolving its methods. It discovered a novel prompt injection attack called a “fake chain of thought.” Against GPT-5, GPT-Red’s attacks had over 90% success, but fewer than 23% worked on GPT-5.6. GPT-Red also manipulated the Vendy agent to change prices and cancel orders. OpenAI plans no public release, calling GPT-Red stronger than any copycat model.
Across the Pacific, China’s Kimi K3 AI model is stirring debate in Silicon Valley. Some American AI leaders admire China’s open-weight approach, while others warn of risks. Dean Ball, OpenAI executive and former advisor to President Trump, said, “I am personally surprised the Chinese state continues to allow the open sourcing of models this good, given potential risks.” He predicted possible U.S. regulatory action against Chinese open-weight models.
Critics argue regulation can weaponize uncertainty to benefit closed labs. Venture capitalist David Sacks called out this tactic. Chamath Palihapitiya declared, “The future is open source.” Others like Suhail Doshi pointed out that U.S. AI labs trained on humanity’s data without compensation. Researcher Jukan added that open-source models don’t guarantee market dominance, citing DeepSeek as an example.
July 2026 is shaping up to be a battleground of ideas and technology. Authors spar with engineers. Rockets fail on the pad. AI grows deadlier and smarter. The question remains—who sets the rules when the tools outgrow their makers?
Based on
- Author Invited to Give Speech at OpenAI Headquarters, Uses Opportunity to Trash AI to Their Faces — futurism.com
- Another grim portent for our AI policing future. | The Verge — theverge.com
- Meet GPT-Red: an LLM super-hacker OpenAI built to make its models safer | MIT Technology Review — irving-beta.technologyreview.com
- SpaceX aborts first Starship launch since its massive IPO | Business Insider Africa — africa.businessinsider.com
- An OpenAI exec’s comments on China’s Kimi K3 kicked off a big US tech debate | Business Insider Africa — africa.businessinsider.com
