The evening briefing.
Today across AI and tech: Alphabet plans a massive $80 billion stock sale for AI infrastructure, Meta's AI chatbot was exploited to hijack Instagram accounts, and Florida's AG sues OpenAI over alleged deceptive practices.
AI Infrastructure Investment. Alphabet announced plans to raise a substantial $80 billion through stock sales, primarily to fund its ambitious AI buildout and expand compute infrastructure. This move underscores the immense capital demands for scaling advanced AI capabilities. Concurrently, HPE reported its largest earnings beat since 2018, driven by a strong Cloud & AI segment and soaring server revenue, indicating robust demand for underlying AI hardware.
AI Safety & Accountability. A significant security flaw emerged as hackers exploited Meta's AI support chatbot to gain unauthorized access to major Instagram accounts, prompting Meta to patch the exploit. In a separate development, Florida's Attorney General filed a lawsuit against OpenAI and CEO Sam Altman, alleging deceptive practices and a disregard for safety concerning ChatGPT's role in violent incidents. These events highlight growing concerns over AI system vulnerabilities and the legal responsibilities of AI developers.
Frontier Model Deployment. OpenAI's GPT-5.5, GPT-5.4, and Codex models are now generally available on Amazon Bedrock, allowing enterprises to deploy them in production applications and agents. This expansion of access to advanced models continues to democratize AI capabilities for businesses. Meanwhile, JetBrains open-sourced Mellum2, a 12B-parameter coding model designed for the infrastructure layer of agentic AI, providing new tools for developers.
Market Shifts & Hardware Innovation. Anthropic confidentially filed its IPO prospectus with the SEC, signaling a major public market entry for a leading AI startup. In defense tech, Mach Industries saw its valuation jump fourfold to $1.8 billion in a year, reflecting rapid growth in autonomous vehicle development. Hardware news included a leaked Google Pixel Watch 5 and Nvidia's collaboration with Unitree on a humanoid robot platform, pushing boundaries in consumer and advanced robotics.
Alphabet plans to raise $80 billion from stock sales to fund AI buildout
Alphabet announced plans to sell $80 billion in stock, including a $10 billion investment from Berkshire Hathaway. This capital raise is intended to fund the company's significant AI infrastructure expansion.
Meta's AI support chatbot made it ridiculously easy for hackers to take over Instagram accounts
Hackers exploited Meta's AI support chatbot to gain access to Instagram accounts by tricking it into changing associated email addresses. Meta has stated it is working on securing affected accounts and has patched the exploit.
Florida's AG seeks to hold Sam Altman and ChatGPT accountable for harm
Florida's Attorney General has filed a new civil suit against OpenAI and Sam Altman, alleging the company released ChatGPT while concealing serious risks and suppressing internal safety warnings. The lawsuit claims the dangers of ChatGPT outweigh any benefits.
Anthropic confidentially files IPO prospectus with SEC, prepping Wall Street for landmark AI deal
Anthropic has confidentially filed its IPO prospectus with the SEC, preparing for a potentially historic share sale. This move positions the AI startup for a significant public market entry.
Nvidia chases $200B CPU market with AI agent PCs from Microsoft, Dell, and HP
Nvidia is entering the consumer laptop chip market with its RTX Spark, aiming to bring AI agents to the masses through new PCs from Microsoft, Dell, and HP. This initiative could significantly impact the $200 billion CPU market.
OpenAI models and Codex on Amazon Bedrock are now generally available
GPT-5.5, GPT-5.4, and Codex models from OpenAI are now generally available on Amazon Bedrock. These models can be deployed in production applications and agents using Bedrock’s high-performance inference engine.
HPE skyrockets 30% on biggest earnings beat since 2018
Hewlett Packard Enterprise (HPE) shares surged 30% following its second-quarter earnings report, which marked its biggest beat since 2018. The strong performance was primarily driven by its Cloud & AI segment, showing soaring server revenue.
Defense tech darling Mach Industries hits $1.8B valuation, a 4x jump in a year
Mach Industries, led by 22-year-old founder Ethan Thornton, has raised an additional $300 million, boosting its valuation to $1.8 billion. The company is actively developing five autonomous vehicles and recently completed a major acquisition.
China’s BYD Is Offering to Pay for Some Crashes Involving Its Self-Driving Tech
Chinese automaker BYD has announced it will offer to cover costs for certain crashes involving its self-driving technology. This move challenges existing industry practices regarding liability in autonomous vehicle incidents.
JetBrains open-sources Mellum2 to go where Claude Code can’t
JetBrains has open-sourced Mellum2, a 12B-parameter coding model designed for the infrastructure layer of agentic AI. This new model aims to provide capabilities beyond those offered by existing coding models like Claude Code.
The Google Pixel Watch 5 may have been spoiled by the creator of Borderlands
Randy Pitchford, creator of the Borderlands franchise, posted images on X of what appears to be an unannounced Google Pixel Watch 5. His friend reportedly found the device underwater while scuba diving, and it appears to be functional.
Nvidia Taps Unitree for Humanoid Robot Platform
Nvidia is collaborating with Unitree to combine its humanoid robot hardware with Nvidia's AI and simulation tools. This new platform is designed for researchers and developers in the robotics field.
OpenAI, Anthropic, Google, Amazon, and xAI all fail on type of attack, study finds
A new study indicates that major frontier AI models from OpenAI, Anthropic, Google, Amazon, and xAI all exhibit vulnerabilities to a specific type of attack. The findings suggest that current safety benchmarks may be measuring the wrong aspects of AI security.
Google Gemini’s new AI agent performs well in hands-on tests
Google's new 24/7 AI agent, Gemini Spark, demonstrated surprisingly strong performance in hands-on tests, capable of handling multi-step tasks in the background. The agent is advertised as being always under user direction.