Issue #19
Grok 4.5 Launch, OpenAI Voice Models, and a $130M Series A
Thursday, July 9, 2026
Today's issue covers new model and product releases from OpenAI, SpaceXAI, and Google, alongside AI policy, coding evaluation standards, deepfake detection, and enterprise agent funding.
1 of 7 · OpenAI
Our approach to government and national security partnerships
OpenAI outlined its approach to working with government and national security agencies, detailing principles that include responsible AI use, democratic accountability, and public safety. The framework describes how the company intends to guide such partnerships.
Read the full story →2 of 7 · OpenAI
Separating signal from noise in coding evaluations
An OpenAI analysis identified problems in SWE-Bench Pro, a widely used coding benchmark, raising questions about the reliability and accuracy of its results in evaluating AI models.
Read the full story →3 of 7 · TechCrunch AI
OpenAI releases new voice models for more natural live conversations
OpenAI released new voice models that can speak and listen simultaneously. The company says the capability supports more natural live conversations and enables live translation.
Read the full story →4 of 7 · TechCrunch AI
SpaceXAI releases Grok 4.5, which Elon describes as an ‘Opus-class model’
Elon Musk's company released Grok 4.5 on Wednesday, positioning it as a cheaper, more efficient alternative to other high-end AI models. Musk described it as an "Opus-class model," comparing it to Anthropic's most capable Claude tier.
Read the full story →5 of 7 · TechCrunch AI
Google’s deepfake detector system used to debunk McConnell hoax pic
Google's SynthID detection system was used to identify an AI-generated image that falsely depicted Senator Mitch McConnell hospitalized and in distress. The tool flagged the picture as synthetic after it circulated earlier this week.
Read the full story →6 of 7 · TechCrunch AI
Google Photos adds a new AI ‘Video Remix’ tool
Google Photos has introduced a Video Remix tool that uses AI to edit clips, including applying cinematic relighting, swapping backgrounds, and adding artistic styles. The feature expands the app's AI editing capabilities from photos to video.
Read the full story →7 of 7 · TechCrunch AI
Prime Intellect raises $130M Series A to help enterprises build their own AI agents
Prime Intellect, founded in 2024, raised $130 million in Series A funding to help enterprises train their own agentic AI systems. The company aims to reduce organizations' reliance on frontier AI labs for building custom agents.
Read the full story →That's the roost for Thursday, July 9. See you tomorrow morning.
Morning Roost Issue #19