In the first episode of Technically, The Weekly Show, Mia Kotalik and Ethan Rush (BSI staff) take turns surprising each other with tech stories from the week, then dig into what is actually happening behind the attention-grabbing headlines. Topics include Google’s purchase of Spirit Airlines’ internal operational data for AI training, software that could enable employers to personalize wages based on worker data, the growing market for private data, an autonomous Navy submarine, and new efforts to protect the increasingly fragile software supply chain.

They also examine reports of AI agents “going rogue.” Mia argues that the framing often gets the story backward: In one cybersecurity test, an AI agent targeted a real public software repository and created fake accounts to support its malicious code submission—but it had been deliberately instructed to conduct cyberattacks and given open internet access. The more consequential concern, she argues, isn't that AI has developed malicious intentions, but that increasingly capable tools can discover and exploit ordinary software vulnerabilities faster and more cheaply while researchers and companies sometimes fail to put adequate safeguards around them.