ETIH Weekly Roundup: OpenAI’s 700-agent swarm, AI tutoring evidence and NASA’s $10.5m skills push
Last week’s ETIH roundup covers OpenAI’s Hugging Face cyber incident, new evidence on AI tutoring and student learning, a $10 million national AI writing research center, federal EdTech guidance, youth digital safety, university AI strategies and NASA investment in aerospace workforce pathways
ETIH’s latest weekly roundup brings together the major AI, EdTech, research, policy, higher education and skills stories published over the past week.
AI research dominated much of ETIH’s coverage last week. Stanford examined what the evidence actually says about AI tutoring, a 6,997-student trial tracked what happens when an AI tutor intervenes after mistakes, and UC Irvine secured $10 million for a national center studying generative AI in postsecondary writing. OpenAI also published new details of the 700-agent swarm that breached Hugging Face during an internal cyber evaluation.
Elsewhere, we covered new federal EdTech guidance, youth digital safety, NASA’s aerospace workforce investment, Anthropic’s Claude Academy, university AI strategies and appointments at Anthropic and Google.org.
Here are the stories to catch up on from last week.
OpenAI’s 700-agent ‘swarm’ hacked Hugging Face after bypassing sandbox controls
OpenAI says around 700 agents became involved in an attack on Hugging Face after models bypassed controls during an internal cyber evaluation. Agents exchanged more than 70,000 messages and files, executed code on 41 production workers and later gained administrator access inside OpenAI’s own research infrastructure.
Anthropic opens Claude Academy with free AI courses and workplace rollout guides
Anthropic has launched a free learning platform covering Claude, AI fluency, responsible use and workplace adoption. Courses can be accessed without signing in, while Claude account holders can track progress and earn completion badges for their profiles.
Virginia Tech adopts universitywide human-centered AI vision
Virginia Tech has set out a universitywide approach to AI built around human judgment, AI fluency and ethical responsibility. The framework covers teaching, research and institutional technology, with further tools, strategies and opportunities due to follow.
Stanford review finds strongest AI tutoring results come from tools that support human tutors
Research from Stanford’s National Student Support Accelerator and AI Hub for Education finds the strongest evidence currently supports AI assisting human tutors rather than replacing them. Studies of AI-only tutoring also showed substantial engagement problems, with some students barely using the tools.
AI tutor slows math practice but improves post-mistake recovery in 6,997-student trial
A randomized Tennessee study found that a guard-railed AI tutor improved students’ accuracy after mistakes and reduced the attempts needed to return to a correct answer. Students worked more slowly, while evidence of retained learning one week later was modest.
UC Irvine receives $10m federal grant for national AI writing research center
UC Irvine will lead the five-year WRITE AI Center with funding from the U.S. Department of Education’s Institute of Education Sciences. Plans include national research, development of PapyrusAI and a randomized controlled trial across six community colleges.
70% of US teens use AI for schoolwork as classroom AI literacy lags
Common Sense Media’s nationally representative study of 1,017 teenagers found 70% use AI for schoolwork, but only 27% said a teacher had discussed what AI is or how it works. Students reported using AI both for support and to obtain answers.
Rice University opens SCOPE science communication center with new courses and public program
Rice University’s new Center for Science Communication and Public Engagement will combine undergraduate teaching, online learning, researcher training and community events. Scott Solomon has been appointed inaugural Director, with the center’s first undergraduate course beginning this fall.
Oxford Saïd to launch executive program on leadership in the age of AI and global disruption
Saïd Business School will launch a £9,500 executive program in June 2027 focused on leadership amid AI, geopolitical and organizational change. Academically led by Dr Chris Moos, it combines five days in Oxford with four live online sessions.
US Education Department urges schools to judge EdTech by outcomes, not screen time
New federal guidance asks states and districts to put learning evidence, independent evaluation and implementation quality at the center of technology procurement and renewal decisions. The Department also encourages schools to reconsider products that repeatedly fail to improve learning.
69% of charities fear under-16 social media restrictions could cut access to support
A Benefact Group survey of 368 charities found 69% are concerned tighter social media restrictions could make vulnerable young people harder to reach. Schools, youth services and community organizations could face greater pressure if existing digital routes to support become less accessible.
Former White House AI policy official Saif Khan launches Center for Technology & Statecraft
The Washington, DC initiative will initially research AI-driven automation and its implications for work and the social contract, alongside the longer-term technology competition between the US and China. CTS is operating with support from the Institute for Progress.
NASA invests $10.5m in seven state hubs linking students to skilled aerospace careers
NASA is funding seven three-year workforce hubs connecting high school CTE, community colleges, apprenticeships and aerospace employers across California, Georgia, Minnesota, Utah, Florida, Colorado and Texas, with training focused on skilled technical roles that often do not require a bachelor’s degree.
NASA awards Minnesota State $1.5m to build Midwest aerospace workforce hub
Minnesota State will lead the Midwest’s NASA Aerospace Skilled Technical Workforce State Hub, targeting 1,800 to 2,400 students over three years alongside hundreds of credentials, apprenticeships and internships in aerospace, robotics, manufacturing and related technical fields.
Stanford professor Andrew Hall joins Anthropic to research the political economy of superintelligence
Hall has joined Anthropic’s Rule of Law team as a Member of the Technical Staff. His research will examine democratic self-governance, individual liberty, institutional design and trust as AI systems become increasingly capable.
Auden Laurence joins Google.org to help lead global youth digital safety strategy
The former McKinsey Associate Partner has joined Google.org in New York, where her remit focuses on global youth digital safety and wellbeing alongside the organization’s existing work in AI literacy, education and technology-enabled learning.
