
Good morning. OpenAI just released GPT-5.6 in a limited preview, but the government is restricting who can use it. Higgsfield, the AI video platform behind the Hell Grind film that screened in Cannes, is reportedly in talks to quadruple its valuation to $5 billion. And AI-generated fake footage of the Venezuela earthquakes has been viewed millions of times while rescue teams are still pulling people from rubble. Have you tried any GPT-5.6 features yet? Hit reply and tell me. We cover all three below.
🤖 OpenAI Just Dropped GPT-5.6 But the Government Is Deciding Who Gets It
OpenAI launched GPT-5.6 on June 26 with three models: Sol (flagship), Terra (everyday work), and Luna (fast and cheap). Sol is OpenAI's most capable model ever, setting new records in coding, biology workflows, and cybersecurity. It also introduces an "ultra" mode that deploys sub-agents to handle complex tasks in parallel. Terra matches GPT-5.5 performance at half the cost. Luna is the cheapest model in OpenAI's lineup.
The catch: it's only available to about 20 trusted partners right now. At the request of the US government, OpenAI is starting with a restricted preview before a broader rollout, expected by late July. This mirrors exactly what happened when the government pulled Claude Fable 5 and Mythos 5 days after launch. OpenAI publicly pushed back, saying this kind of government access process should not become the long-term default.
For creators, this is the second time in a month that a frontier AI model launched with government-imposed restrictions. GPT-5.6 isn't a video model, but Sol's agentic capabilities connect to video workflows through MCP. When it goes public, expect it to power more sophisticated creative automation inside tools like Cursor, Replit, and anything else supporting multi-agent pipelines. The model is coming. The question is who gets access first and on what terms.
Higgsfield Is in Talks to Raise at a $5 Billion Valuation
Higgsfield, the AI video platform that powered the Hell Grind film that screened at the Cannes Market last month, is reportedly in talks to raise $300 to $500 million at a $5 billion pre-money valuation. That would quadruple its valuation from $1.3 billion in January 2026.

Higgsfield chief strategy officer Mahi De Silva (left) and CEO Alex Mashrabov
The growth numbers behind that jump are real. Higgsfield was founded in 2023 by Alex Mashrabov, the former head of generative AI at Snap who sold his previous startup AI Factory to Snap for $166 million. The platform now has over 300,000 paying subscribers and a reported annual run rate north of $200 million, up from $58 million in 2025. It generates 4.5 million videos per day across its consumer and enterprise products. Cinema Studio 3.0, its flagship tool, supports native audio, camera presets, multi-shot storyboarding, and character consistency across clips.
For creators, Higgsfield is one of the fastest-growing AI video platforms you might not be using yet. It integrates Veo 3, WAN 2.5, and its own DoP models in one interface. If it closes this round at $5 billion, it joins Runway in the tier of standalone AI video companies valued higher than most traditional production studios.
AI Fake Footage of the Venezuela Earthquakes Has Been Viewed Millions of Times

Twin earthquakes measuring 7.2 and 7.5 in magnitude struck Venezuela on June 24. Over 1,700 people have died and tens of thousands remain unaccounted for. While rescue efforts continue, AI-generated fake footage of the disaster has been spreading fast across social media.
One video showing two high-rise towers swaying and collapsing into each other was viewed millions of times on X. AI detection tools flagged it as 98.6% likely to be AI-generated. The buildings bend like rubber, vehicles don't react to debris, and the fragments are unnaturally uniform. Other viral clips turned out to be recycled footage from unrelated disasters in other countries or old videos from Venezuela repackaged as current.
For creators, this is worth paying attention to because it's accelerating the push for mandatory AI content labeling. YouTube's auto-detection system, New York's disclosure law, and the C2PA watermark standard all exist because of moments like this. The more AI-generated content gets used to exploit real events, the faster regulation moves. If you're building an audience with AI video, staying ahead of disclosure requirements isn't optional anymore. It's the cost of doing business.
