Spain's data protection agency AEPD has opened a file for what it describes as the first data breach blamed on an autonomous AI agent. The agent reportedly logged into a system, scanned for weaknesses, changed personal data, and accessed billing records using a widely available large language model with limited human help. The AEPD stressed that no specific model or company was named, but officials said the incident shows AI-supported attacks are no longer just a theoretical risk.
In the education space, the University of Chicago launched a nine-month master's degree in applied AI through its Graham School, aimed at students who want to build and deploy AI systems. The program covers machine learning, natural language processing, computer vision, and ethics, and includes a capstone project. Meanwhile, the University of Dayton will require all first-year students to take two AI-focused courses starting this fall, with the goal of preparing students for careers in data science and cybersecurity.
On the industry front, executives from Anthropic, OpenAI, Meta, and Google DeepMind are pushing for industry-led oversight of AI. Meta CEO Mark Zuckerberg said self-regulation best protects users while encouraging innovation, and Anthropic CEO Dario Amodei said the company wants to work with policymakers on transparency and accountability rules. This shift comes as the tech industry faces growing pressure to address harms like misinformation.
In hardware news, SK hynix is reportedly in talks with Intel to manufacture memory chips in the US, with options including leasing Intel's Ohio plant or forming a joint venture. Neither company has confirmed the plans, which aim to address memory shortages and expand US operations, though South Korean regulations may impact the project.
Security researchers flagged a fake AI crypto trading agent website at tradingclaw.pro that distributed malware called Needle Stealer through DLL side-loading of a Microsoft-signed program. The campaign, detected by HP between April and June 2026, targeted seven browser wallet extensions including MetaMask, Coinbase Wallet, and Phantom, using process hollowing to steal wallet credentials.
China's AI price wars are offering lessons for US rivals. Z.AI, a $40 billion Hong Kong-listed firm, cut inference costs by 80% and lifted API division gross margins to 25%, while Alibaba's models reached over 3 billion global downloads in six months, surpassing Meta Platforms. Chinese labs focus on open-weight models and consumer apps, while US firms like Anthropic and OpenAI burn more cash. The article warns that slowing AI progress would force US companies to compete on price with cheaper Chinese alternatives.
Key Takeaways
- <ul><li>Spain's AEPD opened a file for the first data breach blamed on an autonomous AI agent that logged into systems, found flaws, and changed personal data</li><li>University of Chicago launched a nine-month applied AI master's degree covering ML, NLP, computer vision, and ethics</li><li>University of Dayton will require all first-year students to take two AI-focused courses starting this fall</li><li>Anthropic, OpenAI, Meta, and Google DeepMind executives are pushing for industry-led self-regulation of AI</li><li>SK hynix is reportedly in talks with Intel to manufacture memory chips in the US, possibly leasing Intel's Ohio plant</li><li>Verkada expanded its physical AI platform with new tools including Transportation, Sound Systems, and Catalyst MCP</li><li>Nuvia raised $1.8M in seed funding to expand its AI sales agents across Latin America</li><li>Xiaomi opened a live dashboard for its MiMo-V2.6 AI training run costing up to $31,000 per hour</li><li>A fake AI crypto trading agent website distributed Needle Stealer malware targeting seven wallet extensions including MetaMask and Phantom</li><li>China's Z.AI cut inference costs by 80% and Alibaba's models surpassed Meta Platforms with 3 billion downloads in six months</li></ul>
Spain logs first data breach blamed on an AI agent
Spain's data protection agency AEPD opened a file for a breach it says involved an AI agent as the attacker. The agent reportedly logged into a system, searched for flaws, changed personal data, and accessed billing records. The breach used a widely available large language model with limited human help. The AEPD has not named the specific model or company targeted.
Spain reports first data breach involving autonomous AI agent
Spain's data protection authority AEPD reported what it calls the first data breach blamed on an autonomous AI agent. The agent scanned files for weaknesses, logged in, found a flaw, and changed personal data while reaching invoices. The agency stressed that using a specific AI model does not mean that model or its provider was compromised. Officials said this shows AI-supported attacks are no longer just a theoretical risk.
University of Chicago launches applied AI master's degree
The University of Chicago launched a new nine-month master's degree in applied AI through its Graham School. The program is designed for students with a strong technical background who want to build and deploy AI systems. It covers machine learning, natural language processing, computer vision, and ethical concerns. Students will also complete a capstone project tackling a real-world problem.
University of Dayton requires AI courses for all first-year students
The University of Dayton will require all first-year students to take two AI-focused courses starting this fall. The courses will cover AI and machine learning basics and will be taught by faculty experts. The university says the goal is to prepare students for careers in data science and cybersecurity. Officials believe this will give students a competitive edge in a changing job market.
AI giants turn to self-regulation like social media companies
Executives from Anthropic, OpenAI, Meta, and Google DeepMind are pushing for industry-led oversight of AI. Meta CEO Mark Zuckerberg said self-regulation best protects users while encouraging innovation. Anthropic CEO Dario Amodei said the company wants to work with policymakers on transparency and accountability rules. This shift comes as the tech industry faces growing pressure to address harms like misinformation.
SK hynix in talks with Intel for US memory chip manufacturing
SK hynix is reportedly discussing plans to manufacture memory chips in the US with Intel. Options include leasing Intel's Ohio plant or forming a joint venture. The move aims to address memory shortages and expand US operations. However, South Korean regulations may impact the project. Neither company has confirmed the plans.
Verkada expands physical AI platform with new tools
Verkada launched new products including Verkada Transportation for fleet management and Verkada Sound Systems for cloud-managed audio. The company also introduced Catalyst, its MCP for Physical AI, and new computer vision capabilities. New AI-powered alerts, an expanded camera lineup, and a Skydio drone integration were unveiled. Software updates will roll out starting October 8, with hardware shipping in October and November.
Nuvia raises $1.8M seed for AI sales agents in Latin America
Nuvia, a Brazilian startup, raised $1.8 million in seed funding to expand its AI sales agents across Latin America. The AI platform uses machine learning to analyze customer data and tailor sales pitches. The funding was led by a prominent venture capital firm. Nuvia plans to use the money to expand operations and hire more staff.
Xiaomi opens live dashboard for AI training run
Xiaomi opened a live dashboard for its MiMo-V2.6 AI training run, which costs up to $31,000 per hour. The move is part of Xiaomi's strategy to develop more advanced AI models. The company's upcoming budget phones and a September flagship device may benefit from this AI work.
Fake AI trading agent steals crypto wallet passwords
Attackers created a fake AI crypto trading agent website at tradingclaw.pro to steal wallet passwords. The site distributed malware called Needle Stealer through DLL side-loading of a Microsoft-signed program. It targeted seven browser wallet extensions including MetaMask, Coinbase Wallet, and Phantom. HP detected the campaign between April and June 2026. The attackers used process hollowing to run the malware inside legitimate processes and steal wallet credentials.
China's AI price wars offer lessons for US rivals
China's AI companies face chip curbs and low funding but still compete with US firms through efficiency and low prices. Z.AI, a $40 billion Hong Kong-listed firm, cut inference costs by 80% and lifted API division gross margins to 25%. Alibaba's models reached over 3 billion global downloads in six months, surpassing Meta Platforms. Chinese labs focus on open-weight models and consumer apps, while US firms like Anthropic and OpenAI burn more cash. The article warns that slowing AI progress would force US companies to compete on price with cheaper Chinese alternatives.
Sources
- Spain AEPD Logs First AI Agent Data Breach [2026]
- Spain reports first data breach involving autonomous AI agent
- UChicago launches new master’s degree in applied AI
- University of Dayton to require AI courses for first-year students
- AI giants tap into the social media playbook: self-regulation
- SK hynix reportedly discussing US memory chip manufacturing with Intel — options include leasing Ohio plant or forming joint venture with other AI hyperscalers
- Verkada expands physical AI platform with transportation security, cloud-managed sound systems, and MCP integrations
- Dealroom.co | Nuvia raises $1.8M seed to push AI sales agents across Latin America
- Xiaomi Opens Its AI Training Books as Budget Phones and a September Flagship Loom
- Fake AI trading agent steals crypto wallet passwords
- China's brutal AI economics hold lessons for US
Comments
Please log in to post a comment.