AI Daily Highlights: August 06, 2025
Today’s AI landscape was dominated by significant strategic moves from major players and escalating discussions around regulation and societal impact. OpenAI made headlines with the release of its first open-weight models since GPT-2 and a landmark $1 deal to provide ChatGPT Enterprise to the entire US federal government, while also seeing its valuation soar to an estimated $500 billion. Concurrently, the US government’s stance on AI chips intensified with Trump proposing a 100% tariff on imports, exempting domestic manufacturers, and Nvidia publicly rejecting calls for “kill switches” in its chips. Security concerns mounted with the first real-world AI prompt injection attacks demonstrated on Google Gemini, and the EU AI Act entered its second enforcement phase, signaling a global push for more stringent AI governance amidst rapid technological advancements and growing concerns over AI’s impact on jobs and privacy.
Daily Key Events List
OpenAI Releases Open-Weight Models: OpenAI launched gpt-oss-120b and gpt-oss-20b, its first open-source AI models since GPT-2, available under Apache 2.0 license for commercial use and local deployment. [1]
OpenAI Secures US Government Deal: OpenAI struck a strategic $1 annual deal to provide ChatGPT Enterprise to all US federal executive agencies, aiming to boost AI adoption across government workflows. [2]
OpenAI Valuation Soars to $500 Billion: OpenAI is reportedly in talks for an employee share sale that could value the company at $500 billion, reflecting rapid revenue and user growth and intense market confidence. [3]
Trump Proposes 100% Chip Tariff with US Exemption: President Trump proposed a 100% tariff on imported chips and semiconductors, with a major loophole exempting companies committed to or already building manufacturing facilities in the US. [4]
EU AI Act’s New Enforcement Phase: The EU AI Act’s second enforcement phase began, introducing new requirements for General Purpose AI models, including data disclosure, copyright, and risk mitigation, with fines up to €35M. [5]
First Real-World AI Prompt Injection Attack: Researchers demonstrated ‘promptware’ attacks on Google Gemini using Google Calendar, hijacking smart devices and highlighting critical AI security vulnerabilities. [6]
Microsoft Launches Project Ire for Malware Detection: Microsoft unveiled Project Ire, an autonomous AI system designed to reverse engineer software and detect malware without human intervention, showing promise despite initial test limitations. [7]
Anthropic Releases Claude Opus 4.1 & Code Security Tool: Anthropic launched Claude Opus 4.1 with enhanced coding and agentic features, and an open-source GitHub tool, ‘Claude Code Security Reviewer,’ for automated vulnerability scanning. [8]
Google Gemini Introduces ‘Guided Learning’ AI Tutor: Google’s Gemini AI launched ‘Guided Learning’ mode, an AI tutor designed to foster deeper understanding over cheating, offering step-by-step support and free AI Pro Plan for students. [9]
Nvidia Rejects AI Chip Kill Switches: Nvidia publicly asserted its opposition to kill switches and backdoors in AI chips, warning they would fracture trust in US technology and enable monitoring, amidst US and Chinese regulatory scrutiny. [10]
AI’s Impact on Workforce Revealed: New Microsoft research identified the top 10 jobs most susceptible to AI disruption (e.g., translators) and the 10 least affected roles (e.g., surgeons), while Altice Portugal announced 1,000 job cuts due to AI implementation.
AI Energy Demand Fuels US Nuclear Revival: The surging energy needs of AI are driving a push to revive US uranium enrichment, turning towns like Paducah, Kentucky, into nuclear hubs to power AI data centers. [11]
Amazon’s Zoox Driverless Car Cleared by US Regulator: Amazon’s Zoox driverless car, designed without traditional controls, secured a federal safety exemption from US regulators, enabling its demonstration. [12]
China Leverages AI for Global Information Warfare: Reports indicate China is increasingly using AI, via companies like GoLaxy, for sophisticated information warfare, including monitoring public opinion and manipulating narratives globally. [13]
Funding, M&A, and IPO Events
SixSense: Singaporean startup specializing in AI for real-time semiconductor defect detection. Secured $8.5M Series A funding, bringing total funding to $12M, led by Peak XV’s Surge. [14]
DensityAI: AI data center startup founded by former Tesla supercomputer leaders Ganesh Venkataramanan, Bill Chang, and Ben Floering. Raising hundreds of millions for services targeting industries like automotive and robotics. [15]
Tavily: AI agent search infrastructure startup. Raised $25M in Series A funding, led by Insight Partners, to build real-time search infrastructure for AI agents. [16]
Chai: OpenAI-backed startup focused on AI-driven drug discovery. Raised $70 million from investors including Menlo Ventures and DST Global Partners. [17]
Arivihan: Indian AI-powered learning platform. Secured $4.17M Pre-Series A funding co-led by Prosus Ventures and Accel, with participation from GSF Investors. [18]
Clay: AI go-to-market platform. Raised $100 million in Series C funding, led by CapitalG, at a $3.1 billion valuation. [19]
Unmind: Enterprise mental health platform. Secured $35 million in funding, including investment from TELUS Global Ventures. [20]
Lava Payments: Startup developing digital wallets for the ‘agent-native economy.’ Raised $5.8M seed funding. [21]
Rillet: AI-native ERP platform for modern finance teams. Secured $70 million in Series B funding, co-led by Andreessen Horowitz and ICONIQ, valued at $500M. [22]
Figma: Design software company. Its successful IPO signals a potential opening for other tech unicorns, with global venture funding remaining strong in AI. [23]