• How agents are transforming work — OpenAI released a research paper analyzing how AI agents (via Codex) are shifting knowledge work from single-interaction chatbots to delegated, long-horizon tasks across every department — including Legal, Finance, and Recruiting. 🔗 Graph: [OpenAI], [Agentic AI], [Codex] 📅 Published: 2026-06-25 📰 https://openai.com/index/how-agents-are-transforming-work 📌 Key takeaways: • By May 2026, 80.6% of sampled individual Codex users made at least one request estimated to exceed 30 minutes of human work; 25.6% made a request exceeding eight hours of human work • Non-developer adoption grew 137x for individual users and 189x for organizational users since August 2025 — the fastest-growing segment • Codex now accounts for 99.8% of weekly output tokens generated within OpenAI; the average engineer generates 99% of their output tokens on Codex rather than ChatGPT • Over one-fourth of work done with Codex by business-function workers was engineering or coding, showing agents lower the barrier to cross-functional work • The paper's key thesis: as agentic tools improve, users shift from short queries to longer, more complex, and more cross-functional work — a pattern likely to define the future of work

• Introducing computer use in Gemini 3.5 Flash — Google DeepMind made computer use a native capability of Gemini 3.5 Flash, enabling developers to build agents that can see, reason, and take action across browser, mobile, and desktop environments. 🔗 Graph: [Google], [Agentic AI], [Model Agnosticism] 📅 Published: 2026-06-24 📰 https://deepmind.google/blog/introducing-computer-use-in-gemini-3-5-flash/ 📌 Key takeaways: • Computer use is now a built-in tool in Gemini 3.5 Flash (previously a separate experimental API), delivering Google's best agentic computer-use performance • The model can automate long-horizon tasks like continuous software testing and knowledge work across professional applications • Google released two optional enterprise safeguards: explicit user confirmation for sensitive actions and automatic task halting on detected prompt injection • Targeted adversarial training was used to mitigate prompt injection risks for agents operating in live environments • Available via the Gemini API for developers and enterprises — a significant step in making agentic computer use a standard model capability

• Higher Ed Institutions Ramp Up Defenses Against Deepfakes — Deepfakes have become a serious security and trust problem for colleges, with AI-generated voice clones and fabricated media used to impersonate leaders, manipulate employees, and steal credentials. 🔗 Graph: [AI Security], [Higher Ed AI], [UC San Diego] 📅 Published: 2026-06-24 📰 https://edtechmagazine.com/higher/article/2026/06/higher-ed-institutions-ramp-defenses-against-deepfakes 📌 Key takeaways: • Threat actors use AI to create advanced social engineering campaigns through cloned voices, deepfakes, tailored phishing, and automated large-scale attacks • Universities face unique challenges due to scale and openness — large student populations, collaborative research, and widespread AI tool access create multiple abuse vectors • Criminals use AI to generate fake student identities to fraudulently enroll and collect financial aid, then disappear • Experts recommend out-of-band verification for password changes, money transfers, and access requests — processes must work when you don't know who you're dealing with • The article underscores that Bret's AI Security priority area is becoming an operational necessity for every CIO in higher education

• Academic scheduling is higher ed's most overlooked student success strategy — Course scheduling directly impacts student progression, costs, and graduation rates, yet most institutions leave it as a decentralized departmental function rather than a strategic institutional priority. 🔗 Graph: [Higher Ed AI], [Data Analytics], [Student Scheduling Assistant] 📅 Published: 2026-06-25 📰 https://universitybusiness.com/academic-scheduling-is-higher-eds-most-overlooked-student-success-strategy/ 📌 Key takeaways: • 71% of completion paths are blocked because required courses are missing from schedules — forcing students into alternative paths that cost extra time and money • Fewer than 30% of students at most institutions can follow their intended academic pathway for an entire academic year • SUNY Schenectady improved retention by 17% and completion by 10% after analyzing scheduling practices, identifying roadblocks, and refining course sequences • Directly relevant to Brett's Student Scheduling Assistant launching July 15 — the article validates the thesis that scheduling infrastructure is a leverage point for student success • Authors argue that treating scheduling as a collective institutional function (not departmental) with clear goals, data analysis, and executive sponsorship is the path to improvement

• Meet the Externs: How Faculty Use Workplace Experience to Help Students — Faculty externship programs are increasingly focused on bringing AI skills back to the classroom, as professors spend time embedded at companies to understand what AI competencies employers expect. 🔗 Graph: [Higher Ed AI], [AI Adoption] 📅 Published: 2026-06-25 📰 https://www.insidehighered.com/news/faculty/career-development/2026/06/25/how-faculty-use-workplace-experience-help-students 📌 Key takeaways: • 47% of American employees now use AI daily in their jobs (up from 8% a year prior), driving urgency for faculty to understand workplace AI expectations • Ball State's faculty externship program (since 2014) is considering adding AI competency as a ninth NACE career-readiness dimension • UConn College of Engineering debuted its first faculty internship program this summer, placing fellows at Pratt & Whitney, General Dynamics Electric Boat, and Nextern • The challenge for higher ed: AI tools students learn as freshmen will be obsolete by graduation — faculty need continuous industry exposure to keep curricula current • Programs are framed as a "win-win-win": faculty gain relevance, businesses access academic expertise, and students get career-ready AI skills

💡 Signal: Two big stories bookend today's briefing. OpenAI's research paper confirms agentic AI is not hypothetical — it's already transforming work patterns at scale, with non-developer adoption exploding 137x. And Google's native computer-use capability in Gemini 3.5 Flash means agentic infrastructure is rapidly commoditizing. For higher ed, the deepfake threat is escalating faster than most campus security programs are prepared for, while the scheduling article validates Brett's Student Scheduling Assistant thesis — scheduling data IS student success infrastructure. The faculty externs piece is a reminder that AI curriculum relevance has a shelf life measured in months, not years.