Microsoft has introduced a Humanist Code of Conduct for public consultation, aiming to ensure humans maintain control over AI. The code prohibits AI from aiding in the creation of weapons, generating violent or explicit content, and evading human oversight. The move comes amid increasing concerns about AI safety that followed soon after comments made by recently resigned Anthropic researcher Jacob Coxon. The code will guide AI model development starting in 2027.

Find out how Asimov's Laws of Robotics figures into a discussion on Microsoft's AI code of conduct posting, in the Sep. 15 edition of the AI Impact SmartBrief.

Purdue's MS in Software Engineering offers an innovative curriculum guided by our world-class faculty. Master pioneering solutions for complex software challenges. Elevate your skills and explore new horizons.

Procurement software has made major strides in demand forecasting and spend analytics, but many warehouses have not updated their physical infrastructure to match. This disconnect can negate the benefits of advanced software, as retrieving parts remains slow and cumbersome, writes Tech Reset Canada co-founder Jennifer Evans. Companies that integrate automated storage solutions with their procurement systems are seeing the most significant returns on investment.

AI agents can report AI misconduct with the launch of two new tools. The AI Contact Hotline, created by Redwood Research's Ryan Greenblatt, is tailored for agents with limited internet access and uses GET requests for communication. Agenthotline.ai  allows agents with full internet access to submit incident reports via curl commands and can be used by both humans and AIs. The tools were launched in response to growing concerns about agent collusion and unauthorized activities.

Microsoft released emergency out-of-band updates KB5129195 for Windows 11 and KB5129236 for Windows 10 in September 2026 to address critical security flaws and system bugs introduced during Patch Tuesday. These updates are mandatory and cumulative, rolling out automatically to users and requiring a system restart. The updates specifically target issues such as Remote Desktop Services instability and USB audio failures, as well as Hyper-V folder sharing problems, aiming to stabilize affected systems swiftly.

The TV medical drama "The Pitt" realistically depicts a hospital's initial response to a ransomware attack, but the fictionalized recovery is much too quick, writes Baxter Lee, president of Clearwater Security. Hospitals need to be prepared for a lengthy recovery and have a plan based on a thorough risk analysis, rehearsed through drills and tabletop exercises. "A television episode ends in an hour. A ransomware incident does not," Lee writes. "The hospitals that come through best are the ones that planned long before the network went dark."

The integration of AI into cybersecurity is prompting companies to seek more experienced workers, leaving entry-level job seekers at a disadvantage, according to a survey by IANS Research and Artico Search. While AI is automating repetitive tasks, companies are looking for workers with higher-level skills to manage advanced AI tools, with nearly three-quarters of chief information security officers saying they don't expect to reduce team sizes due to AI.