Alex Stamos, former Meta security chief, warns that giving AI agents access to financial logins is completely insane. Now serving as chief security officer at AI coding startup Cognition, Stamos highlights the growing concerns around AI security and the risks of entrusting sensitive systems to autonomous tools.
Key Takeaways
- <ul><li>Former Meta security chief Alex Stamos calls AI access to financial systems insane</li><li>Anthropic's internal AI costs now exceed its employee compensation budget</li><li>Nvidia and AMD back $6B AI startup Hark alongside Intel and Salesforce</li><li>OpenAI warns AI could influence the 2028 White House race</li><li>Amazon expands Alexa for Shopping amid rising AI shopping agents</li><li>ChatGPT introduces PIN lock for private conversations</li><li>HP urges businesses to build AI protection into device architecture</li><li>CFA Institute adds AI training to finance exams</li><li>Anthropic files for IPO as AI bubble concerns grow</li><li>Intel drops nearly 30% after earnings focused on CPUs over GPUs</li></ul>
Former Meta security chief calls AI financial access insane
Alex Stamos, former Meta security chief, says giving AI agents access to financial logins is totally insane. He now works as chief security officer for AI coding startup Cognition. His warning highlights growing concerns about AI security risks.
Anthropic's internal AI use costs now exceed employee pay
Anthropic's internal AI usage, valued at retail prices, now exceeds the company's employee compensation. The report shows how much AI tools like Claude cost to run internally. This reveals the growing financial burden of AI operations for companies.
Nvidia and AMD back $6 billion AI startup Hark
Nvidia and AMD are backing Hark, a $6 billion AI startup that competes with Elon Musk's Grok Bot. Hark raised over $700 million in a Series A round led by Parkway Venture Capital. The company is building AI assistants that can manage calendars, order groceries, and handle web tasks. Investors include Intel, Qualcomm, ARK Invest, and Salesforce.
Salesforce backs $700 million AI assistant startup Hark
Salesforce Ventures joined a $700 million funding round for AI startup Hark. Hark is building proprietary AI foundation models to power a personal assistant that rivals Grok. The investment expands Salesforce's AI reach into consumer-facing assistant technology beyond its core CRM products.
OpenAI lobbyist warns AI could shape 2028 White House race
OpenAI's chief global affairs officer Chris Lehane warned 2028 presidential candidates that AI could decide the White House race. He urged candidates to support AI development to keep the US competitive with China. The private memo was sent to about half a dozen prospective candidates, lawmakers, and political operatives.
Amazon expands Alexa for Shopping to compete with AI agents
Amazon announced new Alexa for Shopping capabilities on Oct. 29 to help users find and buy products directly through Alexa. The move comes as AI shopping agents like Muse challenge Amazon's role as a first stop for shoppers. Amazon CEO Andy Jassy said the company is not competing with third-party agents. A recent study found Amazon accounts for 49% of main purchase completion and 59% of AI-assisted purchases.
Verified AI hardware prices from NVIDIA, Microsoft, MSI, ASUS and HPE
A Top 10 AI hardware manufacturers list could not be fully verified by Shattered.io. Only NVIDIA, Microsoft, MSI, ASUS and HPE had official product pages with real prices ranging from $5,499 to $8,499. Companies like AMD, Intel, Google, AWS, SambaNova Systems, Groq and Graphcore lacked verifiable pricing or specifications. NVIDIA CEO Jensen Huang said AI reinvented the computer industry while AMD's Lisa Su stressed that no single chip design fits every workload.
Middleware will determine success of enterprise AI agents
Enterprise AI agents depend heavily on middleware and integrations to connect to systems across workflows. The middle layer acts as connective tissue between reasoning and execution in tools like CRMs and ERPs. Companies using platforms such as Workato, MuleSoft or Apigee will find adding agents easier. AI agents also expose technical debt from disconnected systems and manual processes that previously compensated for data gaps.
CFA Institute updates exams with hands-on skills and AI training
The CFA Institute is revamping its finance exams to include more hands-on skills and AI training. The changes aim to keep the qualification relevant in the digital age for fund managers. The announcement was made on November 2, 2026. The exams have long been the standard for finance professionals.
ChatGPT adds PIN lock for private chats
OpenAI lets ChatGPT users lock private conversations with a four-digit PIN, Face ID or Touch ID. Locked chats can be accessed temporarily and stored in a dedicated Locked chats section. However, locking does not provide end-to-end encryption or a separate encrypted vault. Users cannot archive, pin, move or share a locked conversation. Access ends after one minute of inactivity on mobile devices, and refreshing a web page ends access as well.
HP Highlights Endpoint Security as Key to Trusted AI Adoption
HP says businesses must have visibility into where AI tools operate and what data they access. The company recommends building protection into device architecture from the start using isolation technology. HP believes clear governance and support from channel partners can help organizations move from AI experimentation to responsible use. Wright from HP explains that isolation helps contain risky activity and reduce the attack surface as agentic AI becomes more common.
theCUBE Pod Discusses AI Bubble and Anthropic IPO Filing
The theCUBE Pod covered recent stock market concerns about an AI infrastructure bubble. Stocks for Intel, Alphabet, Amazon, CoreWeave and Fortinet fell as investors questioned heavy AI spending. Anthropic filed for an IPO with the U.S. Securities and Exchange Commission. Analysts Vellante and Furrier discussed how the dot-com bubble compares and said the current AI market shows signs of a healthy correction rather than a crash. Intel dropped nearly 30% after its earnings report due to its focus on CPUs over GPUs.
Sources
- Ex-Meta security chief: Giving AI agents financial logins is 'insane'
- Anthropic's internal AI usage now exceeds employee compensation
- Nvidia and AMD Are Backing a $6 Billion AI Startup Taking on Grok
- Why Is Salesforce (CRM) Backing A $700 Million AI Assistant Startup?
- OpenAI Lobbyist: AI Could Define 2028 White House Race
- Amazon Positions Alexa for Shopping to Counter Third-Party Agents
- Top 10 AI Hardware Manufacturers in the USA [2026]
- The middle layer will make or break enterprise AI agents
- CFA Institute shifts towards hands-on skills and AI training
- ChatGPT now lets you lock private chats with a PIN
- Securing AI At The Edge: Why The Endpoint Is Becoming Critical To Trusted AI Adoption
- On theCUBE Pod: AI bubble lets out some air
Comments
Please log in to post a comment.