OX Security launched OX Cloud on September 16, 2026, a product designed to give security teams real-time visibility into how AI agents operate in production environments. The platform combines traditional CNAPP capabilities with AI Detection and Response (AIDR) and Agentic Attack Surface Management (AASM). It is available immediately to new and existing customers.
Meanwhile, Spain's data protection agency reported what it calls the first personal data breach caused by an AI agent-led attack on September 14. Cisco announced new Splunk capabilities for agentic security operations on September 15, including specialized AI agents for detection, investigation and response. Expereo research found that 58 percent of enterprises say security and privacy concerns remain a barrier to AI adoption.
In Washington, South Carolina lawmakers held the first meeting of a Special Committee on Artificial Intelligence on September 16. Senator Tom Young Jr leads the committee, which will study AI use in government, education and law enforcement over the next two years. Professors discussed generative AI tools including ChatGPT, Claude and Copilot during the hearing.
OpenAI research studying over 1.5 million ChatGPT messages from April to July 2026 found that workers increasingly use AI for tasks outside their normal job descriptions. Cross-occupation tasks grew from 13.1 percent of AI activity in April to 25.9 percent in July, suggesting companies should rethink how they design work alongside AI access.
Experts warn that rushing agentic AI security tools into businesses could create new vulnerabilities. OpenAI president Greg Brockman advised security teams to use tools like Codex with access to company codebases, but critics say this approach is dangerous. Human operators remain the greatest point of failure, according to industry experts.
On the geopolitical front, experts say AI is making it easier for U.S. adversaries like Iran to track American military operations. Iran has used AI tools such as Claude through VPNs to gather public data and manage social media accounts. Anthropic's report highlighted how satellite imagery and public photos are being used to build detailed intelligence on U.S. forces.
Key Takeaways
- <ul><li>OX Security launched OX Cloud with AI Detection and Response (AIDR) and Agentic Attack Surface Management for real-time visibility into AI agent activity</li><li>Spain reported the first personal data breach caused by an AI agent-led attack on September 14, 2026</li><li>Cisco announced new Splunk agentic SOC capabilities with specialized AI agents for detection, investigation and response</li><li>58 percent of enterprises say security and privacy concerns remain a barrier to AI adoption, per Expereo research</li><li>South Carolina formed a Special Committee on AI to study the technology's use in government, education and law enforcement over two years</li><li>OpenAI research found cross-occupation AI tasks grew from 13.1% to 25.9% of ChatGPT activity between April and July 2026</li><li>Experts warn rushing AI security tools into businesses could create new vulnerabilities rather than fix them</li><li>Iran reportedly uses Claude through VPNs to gather intelligence on U.S. military operations</li><li>Anthropic's report highlights how satellite imagery and public photos are being used to build detailed intelligence on U.S. forces</li><li>OpenAI president Greg Brockman advises using tools like Codex with access to company codebases, though critics flag risks</li></ul>
OX Security launches OX Cloud with AI agent runtime defense
OX Security launched OX Cloud on September 16, 2026, a cloud security product built for environments where AI agents access data and take actions autonomously. The platform combines CNAPP capabilities with AI Detection and Response (AIDR) and Agentic Attack Surface Management. It includes Cloud Security Posture Management, Kubernetes Security Posture Management, Data Security Posture Management, runtime vulnerability detection, cloud inventory and graph-based attack path analysis. The product gives security teams real-time visibility into AI agents, models and MCP servers.
AI agents can strengthen an organization's cybersecurity posture
AI agents can help cybersecurity teams handle more threats with less resources by interpreting information and reasoning across multiple steps. They can improve threat detection by correlating data from multiple sources and identifying suspicious patterns like unusual logins or impossible-travel indicators. AI agents can also automate incident response by gathering information, correlating alerts and initiating predefined response workflows such as isolating compromised devices. Additionally, they help prioritize vulnerability remediation based on contextual risk. Organizations should implement least-privilege access, strong authentication and human oversight to address governance challenges.
Cybersecurity roundup: data identity and AI agents are the new battleground
Spain's data protection agency reported the first personal data breach caused by an AI agent-led attack on September 14. Cisco announced new Splunk capabilities for agentic security operations on September 15, including an agentic SOC with specialized AI agents for detection, investigation and response. Expereo research found that 58 percent of enterprises say security and privacy concerns remain a barrier to AI adoption. CISA and NIST also issued new guidance on protecting cloud identity tokens. Together these developments show that controlling data, identity and AI agents is the new cybersecurity battleground.
OX Security launches OX Cloud for AI agent security
OX Security launched OX Cloud on September 16, designed to give security teams real-time visibility into how AI agents operate in production environments. The product adds AI Detection and Response (AIDR) and Agentic Attack Surface Management (AASM) alongside traditional CNAPP features. It detects and governs AI activity across cloud and runtime environments while discovering agents, AI models, MCP servers and non-human identities. OX Cloud is available immediately to new and existing customers and was developed by the New York-based company led by cofounder and CEO Neatsun Ziv.
South Carolina lawmakers hold special committee on AI
South Carolina state senators held a special committee hearing on artificial intelligence on Wednesday at the South Carolina State House in Columbia. The hearing was convened due to growing concerns about possible rogue AI and related issues creating conversation across the country. The full committee hearing was available to watch via video. The event reflects increasing attention from lawmakers on the potential risks and governance of AI technology.
South Carolina Senate committee holds first meeting on AI
South Carolina lawmakers and University of South Carolina professors met on September 16, 2026 to discuss artificial intelligence. Senator Tom Young Jr leads the new Special Committee on Artificial Intelligence, which will study AI use in government, education, and law enforcement over the next two years. Professors Bryant Smith and Homayoun Valafar discussed generative AI tools like ChatGPT, Claude, and Copilot. The committee plans to meet periodically and is seeking public comment on the technology.
Fear of AI says more about humans than machines
An opinion piece argues that human anxiety about artificial intelligence reflects our own nature more than any real threat from the technology. The article notes that humans uniquely fantasize about their own extinction and that fears about AI echo Cold War-era thinking. The author questions why cooperation with China on AI is not being prioritized and suggests people should examine what they are actually afraid of. The piece calls AI a tool that can also be a weapon in the wrong hands.
Rushing AI security tools into businesses creates risks
Experts warn that deploying agentic AI security tools too quickly in businesses could create new vulnerabilities instead of fixing them. OpenAI president Greg Brockman advised security teams to use tools like Codex with access to company codebases, but critics say this is dangerous. Flavio Villanustre of LexisNexis Risk Solutions said OpenAI should help address security problems with higher safety standards. Experts say human operators remain the greatest point of failure and realistic security simulations matter more than untested AI agents.
AI helps adversaries target U.S. military, experts warn
Experts say artificial intelligence is making it easier for U.S. adversaries like Iran to track and understand American military operations. Iran has used AI tools such as Claude through VPNs to gather public data, translate content, and manage social media accounts on platforms like X, TikTok, and Instagram. The Pentagon is considering countermeasures including AI personas that mimic realistic life patterns to hide military activities. Anthropic's report highlighted how satellite imagery and public photos are being used to build detailed intelligence on U.S. forces.
AI dominance is not inevitable, people can push back
A Guardian opinion piece argues that the rise of artificial intelligence is not unavoidable and society can choose better outcomes. The author points to past examples where international agreements and public pressure changed the path of technology, including gene editing and social media regulation. The article calls for independent oversight, regulation, and global agreements on AI development. It also notes that AI models were built using millions of people's art and writing without permission or payment. The author urges people to contact politicians and protest to demand action on AI.
Workers are finding new ways to use AI beyond their job roles
OpenAI research studied over 1.5 million ChatGPT messages from April to July 2026. Workers increasingly use AI for tasks outside their normal job descriptions. These cross-occupation tasks grew from 13.1% of AI activity in April to 25.9% in July. Workers return to certain cross-occupation tasks at high rates, like discussing goods with customers at 54%. The findings suggest companies should rethink how they design work alongside AI access.
Memory shortage forces phone and laptop makers to redesign products
A memory shortage is forcing smaller device makers to change how they build products. Fairphone CEO Raymond van Eck says memory accounts for about 60% of materials cost in phones around $400. SK Hynix CEO Kwak Noh-jung reported similar figures. Companies like Jolla are designing two motherboard versions to switch between chip packages. Framework relies on modular designs and places non-cancellable orders months in advance. Some makers also test incoming chips to verify they are not fake.
AI safety debate intensifies after whistleblower warns of existential risk
AI leaders are divided over safety concerns following whistleblower Dario Coxon, Anthropic's head of product. Coxon warned AI could potentially harm humans by 2030. CEOs of OpenAI and Anthropic called for a slowdown in AI development over safety fears. Chinese authorities, U.S. President, and Nvidia CEO Jensen Huang dismissed these concerns as fearmongering. The debate highlights deep disagreements within the industry about AI regulation and governance.
AI is already changing cricket even without robots bowling yet
AI is reshaping cricket by helping players, coaches, and analysts do more without replacing them. Analyst Freddie Wilde says almost all parts of his daily work now involve AI. AI lets analysts work independently on tasks that once needed data scientists. The technology can help batters find weaknesses and bowlers adjust strategies. Prof Adeela Ahmed Shafi says data can reveal bias in player development pathways. Journalism and media are also being transformed as AI changes how stories are produced and judged.
Sources
- OX Security Unveils CNAPP Platform Paired With AI Agent Runtime Defense
- AI Agents are here to Improve an Organization’s Cybersecurity Posture
- Cybersecurity Roundup: Who Controls the Data, Identity and AI Agents Is the New Battleground
- OX Security Launches OX Cloud for AI Agent Security
- LIVE: SC lawmakers hold special committee on AI
- South Carolina lawmakers, experts talk artificial intelligence. Here’s what happened
- Are we in the AI end times?
- Hastily deployed agentic security is not the answer to enterprise cyber threats
- AI makes it "easier than ever" for adversaries to target U.S. military, experts say
- The dominance of AI is not inevitable. We can choose to change things for the better
- How workers are unlocking new ways of working
- AI-induced Memory shortage is changing how devices are built, Fairphone says memory now 60% of materials cost — smaller laptop and phone makers are redesigning products and have to test for fake chips
- AI leaders clash over safety fears after Anthropic whistleblower says AI could 'kill us all' by 2030
- Robots bowling the perfect doosra are some way off but AI is already reshaping cricket in myriad ways
Comments
Please log in to post a comment.