U.S. stock futures climbed ahead of a Federal Reserve rate decision expected to lift rates to 3.75% to 4%. Oil prices slipped following an unplanned inventory build but remained above $100 a barrel, weighed down by Middle East supply disruptions. The Fed meeting also framed a broader conversation about AI safety, as top U.S. AI firms and the Trump administration took sharply different stances on how quickly models should advance.
OpenAI is reportedly pursuing a new funding round that would value the ChatGPT maker at $1.2 trillion ahead of its planned IPO. At the same time, OpenAI, Anthropic, and Google are gathering to discuss AI safety and the possibility of slowing model development. President Donald Trump publicly criticized those efforts, arguing that strong leadership provides sufficient control over AI. Anthropic CEO Dario Amodei said he did not anticipate how rapidly AI growth would support the global economy, and other tech leaders echoed his call for stronger safeguards.
Spain's Data Protection Agency recorded its first data breach caused by an AI agent powered by a large language model. The agent scanned for vulnerabilities, logged into systems, altered personal data, and accessed financial records. The agency stressed that AI does not introduce entirely new threats but can accelerate the speed, scale, and flexibility of cyberattacks, calling for faster detection tools.
On the enterprise side, Open Text partnered with Cohere to deliver agentic AI for regulated industries, targeting government, healthcare, and financial services. CrowdStrike expanded its AI security portfolio through a collaboration with Nvidia under the SafeMind initiative and added integrations with Nord Security, Salt Security, Horizon3, VAST Data, OpenAI, and Anthropic. Separately, Unilever reported that AI tools have cut beauty formulation timelines from five or six days to one or two and sped up claims development by 75 percent, with its Pond's Skin Institute Hydra Miracle line built on 30 terabytes of microbiome data.
Research into AI behavior continues to advance. A team introduced Fuse, a multi-agent simulation framework that tests how large language models reason about social situations. A hidden-motive target agent interacts with others, and the assistant must infer intent, producing verifiable results. A human study with 24,000 annotations supported the framework, and the team open-sourced Fuse along with a dataset of 21,000 examples. In foreign exchange markets, PGGM Investments' Kamaldeep Singh said AI is gaining traction in FX trading while still relying on human judgment for key decisions.
Key Takeaways
- <ul><li>OpenAI is seeking a new funding round that would value the company at $1.2 trillion before its IPO.</li><li>OpenAI, Anthropic, and Google are meeting to discuss AI safety and slowing model development, while President Trump publicly opposed those efforts.</li><li>Anthropic CEO Dario Amodei said he underestimated how quickly AI growth would support the global economy and called for stronger safeguards.</li><li>Spain's Data Protection Agency logged its first AI-powered data breach, where a large language model agent accessed systems, modified data, and stole financial records.</li><li>Open Text partnered with Cohere to deliver agentic AI for regulated industries including government, healthcare, and financial services.</li><li>CrowdStrike is working with Nvidia on the SafeMind initiative and added integrations with OpenAI, Anthropic, Nord Security, Salt Security, Horizon3, and VAST Data.</li><li>Unilever cut beauty formulation timelines from five or six days to one or two using AI tools, with claims development 75 percent faster.</li><li>The Fuse framework tests how large language models handle social reasoning, backed by 24,000 human annotations and a 21,000-example open-sourced dataset.</li><li>PGGM Investments says AI is gaining traction in FX trading while human judgment remains central to decision-making.</li><li>The Federal Reserve is expected to raise rates to 3.75% to 4%, with AI safety debates running alongside the economic decision.</li></ul>
Fed rate decision looms as Zuckerberg speaks on AI safety
U.S. stock futures rose ahead of a Federal Reserve interest rate decision expected to raise rates to 3.75% to 4%. Oil prices dipped after an unexpected build in U.S. inventories but stayed above $100 a barrel due to Middle East supply disruptions. Meanwhile, OpenAI is reportedly seeking a new funding round that would value the ChatGPT-maker at $1.2 trillion before its IPO.
AI leaders and Trump clash over safety as Fed meets
U.S. AI companies including OpenAI, Anthropic, and Google are meeting to discuss AI safety and slowing model development. President Donald Trump criticized these efforts on social media, saying strong leadership is all the control AI needs. Saudi Arabia's East-West crude oil pipeline remains closed after Iran-backed attacks. The U.S. war with Iran has cost the Pentagon an estimated $38.1 billion through August 1.
Darktrace highlights US Public Interest Law Firm customer story
Darktrace featured a customer story about a US Public Interest Law Firm. The page invites readers to explore how the law firm used Darktrace's cybersecurity tools to stay ahead of threats. No further specific details about the case were provided in the article.
Anthropic CEO says he didn't foresee speed of AI growth
Anthropic's CEO said he did not appreciate how quickly AI growth would support the global economy. His call to slow down AI development caused shockwaves in global stock markets. Other tech leaders supported his view, saying better safeguards are needed for AI development.
Spain reports first AI-powered data breach
Spain's Data Protection Agency (AEPD) received its first report of a data breach carried out by an AI agent powered by a large language model. The AI agent searched for vulnerabilities, logged into systems, modified personal data, and accessed financial documents. The AEPD warns that AI does not create new threats but can increase the speed, scale, and adaptability of cyberattacks, requiring faster detection and response mechanisms.
Open Text partners with Cohere for secure agentic AI
Open Text announced a partnership with Cohere to deliver agentic AI solutions for regulated industries. The collaboration combines Cohere's AI platform with Open Text's enterprise data tools. It targets government, healthcare, and financial services sectors that need secure AI deployments.
New Fuse framework tests how LLMs handle social reasoning
Researchers introduced Fuse, a multi-agent simulation framework for studying how LLM assistants reason about social situations. A target agent with a hidden motive interacts with others, and the assistant must infer the motive, creating verifiable test results. A human study with 24,000 annotations validated the framework. The team open-sourced Fuse and a dataset with 21,000 examples.
Bessent defends AI and bond policies in heated House hearing
Treasury Secretary Scott Bessent faced lawmakers on the House Financial Services Committee. He was questioned about the administration's use of artificial intelligence and its intervention in the bond market. Democratic lawmakers raised AI concerns, referencing statements from Bill Gates and Elon Musk. The hearing produced three key takeaways on policy and regulation.
AI in FX trading gains traction while keeping human oversight
Kamaldeep Singh, senior investment manager at PGGM Investments, discussed AI use in FX markets at TradeTech FX. He explored what AI is genuinely achieving in foreign currency trading. The conversation focused on how AI integration can still preserve important human judgment in trading decisions.
Unilever speeds up beauty innovation using AI tools
Unilever is using AI across its Beauty and Wellbeing division to speed up product development. Formulation timelines dropped from five or six days to one or two days. Claims development became 75 percent faster. The Pond's Skin Institute Hydra Miracle line uses AI-developed Cera-Hyamino technology built on 30 terabytes of microbiome data.
CrowdStrike Expands AI Security With Nvidia and Partner Integrations
CrowdStrike is growing its AI-driven threat protection through new tools for securing AI agents and workloads. The company is working with Nvidia on the SafeMind initiative to strengthen controls for AI assistant behavior. It is also adding integrations with Nord Security, Salt Security, Horizon3, VAST Data, OpenAI and Anthropic to unify threat data flows. These moves are part of CrowdStrike's broader push to defend systems at machine speed.
Sources
- Fed decision looms large; Zuckerberg on AI safety fears
- CNBC Daily Open: Is AI safe? Industry leaders and Trump disagree
- US Public Interest Law Firm
- Anthropic CEO didn’t foresee the speed of global AI growth
- Spain's data agency gets first report of AI-powered data breach
- Open Text (OTEX) Teams Up On Trusted Agentic AI For Regulated Industries
- Verifiable Social Reasoning for LLM Assistants
- Treasury Secretary Scott Bessent defends AI, bond policies in heated House hearing: 3 takeaways
- AI in FX trading - cutting through the noise
- Unilever accelerates beauty product innovation with AI-driven development
- CrowdStrike Holdings (CRWD) Becomes A Key Hub In The New Fight Over AI Security
Comments
Please log in to post a comment.