Microsoft is consolidating its security tools around AI agents. The company launched the Integrated Security Operations Center, or ISOC, inside Microsoft Defender. The move shifts SIEM features from Microsoft Sentinel into the Defender portal. Teams and AI agents now work from one environment, with case management, workbooks and automation playbooks built from plain-language instructions. Microsoft offers more than 500 connectors for external data sources, though ingestion fees may apply. A public preview is open to customers with Microsoft Defender Suite or Microsoft 365 E5 or E7 licenses, and Defender data is kept for 30 days at no extra charge.
Human oversight remains central to the new setup. High-stakes actions still require human approval under Project Perception, and Microsoft stresses that AI agents get the same context and controls as human analysts. Some capabilities, such as user and entity behavior analytics, need extra setup and an Azure-linked workspace. The goal is to give security teams and AI agents a shared workspace rather than separate tools.
In healthcare, AI is driving up costs in a different way. A Blue Cross Blue Shield Association study found that AI tools added nearly $1 billion in spending over two years. Providers billed more often for secondary conditions between 2024 and 2025, adding $653 million in costs. Overall, more intense care contributed $942 million compared with 2023. Insurers said AI tools identified more billable conditions from patient records, not sicker patients, and the study noted an average increase of nearly $12,000 per case.
Hospital and insurer AI systems are now at the center of billing disputes. Lawmakers point fingers at both sides as claims grow more complex. The dispute highlights how the same technology can push costs up when used in competing workflows.
Amazon Web Services is taking a different approach to AI agents. It published a guide for building multi-account agents using AgentCore Gateway and MCP. The design keeps each team's data in its own AWS account and connects them through a single MCP endpoint. Amazon Bedrock handles AI inference, and tools like Okta support authentication. The setup lets agents query data across accounts without copying or centralizing it.
Regulation and education are also moving forward. The EU proposed the KIDS Act to strengthen child safety rules in AI, while AI CEOs called for global regulation at a UN meeting. Microsoft drafted its own AI code of conduct for its models. Minnesota State Moorhead launched a new master's program in Artificial Intelligence to meet growing demand. At the same time, New Yorker journalists said the public excitement around AI has faded.
On the business side, Artificial Intelligence Technology Solutions raised prices on its security hardware, including AI-powered cameras and software, citing rising production costs. The International Monetary Fund warned that debt-funded AI investments could trigger an equity market selloff, noting the global AI market is expected to reach $1.4 trillion by 2028. Factory's Legacy-Bench showed that frontier models still struggle with older coding systems, with GPT-5.6 Sol leading at 60.0 percent, ahead of Claude Opus 5 at 52.0 percent and GPT-6 Astra at 51.0 percent.
Key Takeaways
- <ul><li>Microsoft launched ISOC inside Defender, moving Sentinel's SIEM features into one workspace for teams and AI agents.</li><li>Human approval is still required for high-stakes actions under Project Perception, keeping oversight central.</li><li>A Blue Cross study found AI tools added nearly $1 billion in costs over two years, including $653 million from extra billing.</li><li>Amazon Web Services published a guide for multi-account AI agents using AgentCore Gateway and MCP.</li><li>The EU proposed the KIDS Act for child safety in AI, and Microsoft drafted its own AI code of conduct.</li><li>Minnesota State Moorhead launched a new master's program in Artificial Intelligence.</li><li>The IMF warned that debt-funded AI investments could hurt equity markets as the global AI market heads toward $1.4 trillion by 2028.</li><li>Factory's Legacy-Bench showed GPT-5.6 Sol leading legacy coding tasks at 60.0 percent, ahead of Claude Opus 5 and GPT-6 Astra.</li><li>New Yorker journalists said the public honeymoon phase for AI is over.</li><li>AITX raised prices on its AI-powered security hardware due to higher production costs.</li></ul>
Microsoft launches ISOC in Defender for AI agents
Microsoft unveiled the Integrated Security Operations Center in Microsoft Defender on September 23, 2026. ISOC moves security information and event management features from Microsoft Sentinel into Defender. It is designed to help security teams and AI agents work from the same environment. The public preview is open to customers with Microsoft Defender Suite, Microsoft 365 E5 or E7 licenses. Defender data is kept for 30 days at no extra charge during the preview.
Microsoft unites Sentinel and Defender for AI agents
Microsoft is bringing security features in Defender closer together so teams and AI agents can work from one environment. SIEM features from Sentinel are now directly available in the Defender portal. Case management, workbooks and automation playbooks built from plain-language instructions work without extra setup. More than 500 connectors are available for external data sources, though ingestion fees may apply. Human approval is still required for high-stakes actions under Project Perception.
Microsoft deepens agentic AI in Defender with ISOC
Microsoft introduced the Integrated Security Operations Center within Microsoft Defender on September 24, 2026. ISOC centralizes security information and event management features from Microsoft Sentinel and rebuilds security operations around AI agents. Some capabilities like user and entity behavior analytics need extra setup and an Azure-linked workspace. Microsoft offers over 500 connectors for external data sources, though ingestion charges may apply. AI agents get the same context and controls as human analysts, but human oversight remains important for strategic decisions.
AI tools added nearly $1 billion in costs for Blue Cross insurers
A Blue Cross Blue Shield Association study released September 24, 2026 found AI tools raised health spending by nearly $1 billion over two years. Providers more often billed for secondary conditions between 2024 and 2025, adding $653 million in costs. Overall, more intense care contributed $942 million compared to 2023. Insurers said AI tools identified more billable conditions from patient records, not sicker patients. The study focused on billing in inpatient and hospital settings.
Hospital AI versus insurer AI is driving up medical costs
A Blue Cross Blue Shield Association report shows hospitals are using AI tools to document more complex patient conditions and submit bigger claims. Insurers found patients were described as having more complicated conditions in 2024 and 2025 than in 2023, despite no evidence of different treatment. This added an average of nearly $12,000 per case. AI has become a key tool in billing disputes between hospitals and insurers. Lawmakers are pointing fingers at each other for rising health costs.
New Yorker journalists say AI excitement has faded
New Yorker journalists Jane Mayer, Susan B. Glasser and Evan Osnos spoke at the University of Wisconsin-Madison. They recorded The New Yorker's The Political Scene podcast during Cap Times Idea Fest. The journalists said the honeymoon phase for AI is over.
AITX subsidiary raises prices on security products
Artificial Intelligence Technology Solutions Inc and its subsidiary Artificial Intelligence Technology Solutions - Co announced a hardware price increase. The price increase applies to all products in their security solution lineup. This includes AI-powered security cameras and software. The company said rising production costs drove the decision. Existing customers will keep the same service and support.
Minnesota State Moorhead adds master's in AI
Minnesota State Moorhead has launched a new master's program in Artificial Intelligence. The program was announced by Valley News Live. It offers students the chance to study AI at the graduate level. The university expanded its offerings to meet growing demand for AI education.
IMF warns debt-funded AI investments could hurt markets
The International Monetary Fund warned that debt-financed AI investments could cause an equity market selloff. The global AI market is expected to reach $1.4 trillion by 2028. Many companies are using debt to fund their AI projects. The IMF says heavy debt loads from these investments could become hard to repay and hurt the global economy.
AWS shows how to build multi-account AI agents
Amazon Web Services published a guide for building multi-account AI agents using AgentCore Gateway and MCP. The architecture keeps each team's data in its own AWS account. AgentCore Gateway connects all accounts through a single MCP endpoint. The system uses Amazon Bedrock for AI inference and supports tools like Okta for authentication. This design lets agents query data across accounts without copying or centralizing it.
EU proposes KIDS Act and AI leaders call for global regulation
The EU proposed the KIDS Act to strengthen child safety rules in artificial intelligence. AI CEOs called for global AI regulation at the UN meeting. Microsoft also drafted an AI code of conduct for its own models. These updates highlight growing efforts to regulate AI more carefully worldwide.
Factory benchmarks AI models on legacy coding systems
Factory's Legacy-Bench tests how well frontier AI models handle software written in COBOL, Java 7, BASIC, C89, Fortran, and Assembly. GPT-5.6 Sol scored 60.0 percent, leading Claude Opus 5 at 52.0 percent and GPT-6 Astra at 51.0 percent. The benchmark covers tasks in financial settlement, payroll, insurance, telecom, and scientific computing. Results show that strong performance on modern coding tasks does not always transfer to legacy systems.
Sources
- Microsoft unveils Integrated Security Operations Center in Defender for AI agents
- Microsoft brings Sentinel and Defender together for AI agents
- Microsoft brings agentic AI deeper into Defender with new ISOC
- AI tools generated nearly $1 billion in extra costs, Blue Cross insurers say
- Battle of Hospital A.I. vs. Insurer A.I. Is Pushing Medical Costs Higher
- New Yorker journalists in Madison: 'The honeymoon is over on AI.”
- Artificial Intelligence Technology Solutions - Co, Its Subsidiary Announce Hardware Price Increase Across Security Solution Lineup
- Minnesota State Moorhead launches master’s program in Artificial Intelligence
- Debt-Financed AI Investment Boom Risks Equity Market Selloff, IMF Says
- Build a multi-account AI agent with AgentCore Gateway and MCP
- Artificial intelligence: key updates and developments (10
- Which model is best for legacy code?
Comments
Please log in to post a comment.