🚪OAI's robotics lead exits over Pentagon deal
OpenAI robotics director Caitlin Kalinowski has publically resigned over the company’s controversial Pentagon deal, stating the agreement was rushed and bypassed necessary safety guardrails.
- —Kalinowski joined OpenAI in November to rebuild its robotics division
- —She emphasized that the decision was based on principle, deeply concerned by the lack of clear guardrails for AI in warfare
- —This marks the first major public resignation specifically tied to the Pentagon controversy, following recent consumer backlash
Why it matters: Enduring consumer boycotts is one thing; losing senior leadership over foundational ethical issues like "lethal autonomy" delivers a distinctly sharper blow to OpenAI’s long-term reputation.
💡The Rundown Roundtable: Our AI use cases
The Rundown Roundtable highlights practical, real-world applications of AI being utilized by the company’s own staff across various operations.
- —Growth team member Rishi shared how he combined Microsoft Clarity data with Claude Code to generate incredibly deep A/B testing ideas
- —The AI provided five prioritized landing page tests alongside their underlying logical rationale, with one test already fully deployed
- —Meanwhile, developer Adrian humorously revealed his usage of Nano Banana 2 to virtually try on new hairstyles
Why it matters: Sharing actionable AI workflows across different organizational roles rapidly accelerates overall AI literacy and inspires teams to automate their tedious, daily tasks.
⚙️ Build an AI case study generator
This step-by-step guide demonstrates how to harness Claude to transform raw project materials—like scattered metrics and client emails—into highly compelling, high-converting professional case studies.
- —Start by compiling a brief project wrap memo containing key KPIs and overall results at the end of an engagement
- —Setup a dedicated "Case Study Generator" project in Claude and initialize it with robust role instructions
- —Feed the raw memo into Claude, instructing it to output a case study utilizing a "challenge-solution-results" framework
Why it matters: Systematizing the extraction of value from disconnected legacy documents creates a powerful sales enablement pipeline, turning forgotten data into proven marketing assets.
🔒Claude digs up 22 Firefox security flaws in two weeks
Anthropic revealed that its Claude Opus 4.6 model spent two weeks auditing the Firefox codebase alongside Mozilla engineers, uncovering 22 significant vulnerabilities, 14 of which were high-severity.
- —Claude flagged its first vulnerability in a mere 20 minutes of scanning
- —In total, Anthropic submitted 112 reports across roughly 6,000 files, resulting in nearly 20% of Firefox’s most severe patches this year
- —Despite its auditing prowess, Claude currently struggles to successfully draft functional exploit code for the bugs it finds
Why it matters: If an AI can rapidly identify two dozen critical vulnerabilities in an ancient, heavily audited open-source project like Firefox, every organization urgently needs to reassess the security of their own proprietary codebases.