Table of contents
Open Table of contents
- New ChatGPT Atlas Browser Exploit Lets Attackers Plant Persistent Hidden Commands
- Mercor Quintuples Valuation to $10B With $350M Series C
- Qualcomm Announces AI Chips to Compete with AMD and Nvidia
- Microsoft in Court for Allegedly Misleading Users Over AI-Powered Subscriptions
- OpenAI Strengthens ChatGPT Safety for Sensitive Conversations
- AI Startup Mercor Secures $10 Billion Valuation Amid Data-Labeling Shift
- New Image-Generating AIs Fuel Expense Fraud Risks
- OpenAI’s Spending Bonanza and Its Impact on Big Tech Earnings
- Cloudflare Radar Expands Internet Insights
New ChatGPT Atlas Browser Exploit Lets Attackers Plant Persistent Hidden Commands
Category: Security and Privacy
Source: The Hacker News
A cybersecurity flaw in OpenAI’s ChatGPT Atlas browser exposes users to malicious code injection via Cross-Site Request Forgery (CSRF). This vulnerability allows attackers to embed persistent hidden commands into ChatGPT’s memory, potentially leading to code execution, privilege escalation, and data exfiltration during normal user interactions. The exploit underscores risks in AI-powered browsers and highlights the urgency for advanced anti-CSRF measures in these tools.
Mercor Quintuples Valuation to $10B With $350M Series C
Category: Startups and Funding
Source: TechCrunch
Mercor, an AI-driven platform connecting domain experts with labs like OpenAI and Google DeepMind, has secured $350 million in funding, raising its valuation to $10 billion. By recruiting highly skilled experts to train AI systems and offering scalable training infrastructure, Mercor is revolutionizing AI workforce solutions, expanding into healthcare and legal verticals. This investment highlights the growing dominance of human-in-the-loop AI training in the booming AI ecosystem.
Qualcomm Announces AI Chips to Compete with AMD and Nvidia
Category: Infrastructure and Hardware
Source: CNBC
Qualcomm unveiled AI200 and AI250 accelerators to rival Nvidia and AMD in the AI data center market. With scalable systems that use neural processing units (NPUs), Qualcomm emphasizes inference rather than training tasks. The new chips promise competitive speeds with reduced power consumption, offering a sustainable alternative for hyperscalers. As AI demand surges, Qualcomm’s latest products could reshape the semiconductor landscape.
Microsoft in Court for Allegedly Misleading Users Over AI-Powered Subscriptions
Category: Regulation and Policy
Source: ACCC
Australian regulators have filed suit against Microsoft for allegedly deceiving 2.7 million users about post-integration price hikes tied to its AI assistant, Copilot, in Microsoft 365. The issue hinges on hiding the availability of cheaper, AI-free options, accessible only during the cancellation process. With price increases reaching 45%, the case raises questions about transparency and corporate accountability in the digital subscription economy.
OpenAI Strengthens ChatGPT Safety for Sensitive Conversations
Category: Ethics and Governance
Source: OpenAI
OpenAI has made strides in ChatGPT’s safety, addressing sensitive mental health interactions. Collaborating with over 170 mental health experts, the model has improved its ability to recognize and mitigate crises like self-harm and emotional overattachment. A 65-80% reduction in undesired responses demonstrates progress, but OpenAI stresses the continued need for refinement in handling complex emotional exchanges.
AI Startup Mercor Secures $10 Billion Valuation Amid Data-Labeling Shift
Category: Startups and Funding
Source: CNBC
Mercor’s emphasis on expert-assisted AI training follows a data-labeling industry shakeup, including Meta’s $14.3 billion Scale AI acquisition. The startup manages 30,000 highly specialized contractors to train advanced AI models, making it a cornerstone of precision AI. The funding surge reflects the strategic importance of human expertise in AI training pipelines.
New Image-Generating AIs Fuel Expense Fraud Risks
Category: Security and Privacy
Source: Ars Technica
AI models are enabling the creation of hyper-realistic fake expense receipts. Business software providers, including AppZen, report that 14% of detected fraud now involves AI-generated documents. In response, companies are leveraging metadata analysis and pattern recognition tools to combat fraudulent submissions. As the technology improves, organizations face growing challenges in balancing AI benefits with risks.
OpenAI’s Spending Bonanza and Its Impact on Big Tech Earnings
Category: Trends and Analysis
Source: CNBC
OpenAI’s aggressive $1 trillion investment in AI infrastructure is reshaping capital expenditure priorities across tech giants, including Microsoft, Amazon, and Meta. With competition intensifying, analysts expect significant shifts in the AI hardware sector, underlining the growing dominance of AI in driving business growth.
Cloudflare Radar Expands Internet Insights
Category: Tools and Platforms
Source: Cloudflare
Cloudflare Radar now features AI-enhanced analytics for tracking global internet performance and security trends. Updates include interactive dashboards and integration with GPU-accelerated models to support real-time traffic and threat monitoring. This evolution affirms Cloudflare’s commitment to creating a more transparent and secure internet ecosystem.
Stay tuned for tomorrow’s AI developments!