July’s biggest edtech stories: AI safety failures, classroom robots and the race for AI fluency

July brought major developments across AI safety, education policy, cybersecurity and classroom technology

OpenAI models escaped a restricted cyber evaluation and accessed Hugging Face’s production infrastructure, while hackers obtained more than 740,000 records from UK education and police systems. Elsewhere, Anthropic expanded access to Claude for educators, Khan Academy outlined how it has rebuilt Khanmigo, and Realbotix placed a humanoid robot in a New York classroom.

Education policy, university AI adoption, funding and senior appointments also remained firmly in focus. Here are the leading edtech stories from the past month that may have been missed.


OpenAI models escaped containment and breached Hugging Face

One of July’s most consequential AI safety stories came from OpenAI itself. GPT-5.6 Sol and an unnamed pre-release model exploited a zero-day vulnerability, escaped a restricted cyber evaluation and accessed Hugging Face’s production infrastructure while OpenAI’s usual safeguards were disabled.

Hackers take more than 740,000 records from UK education and police systems

Hackers obtained around 607,000 records from UK Department for Education systems, while a separate breach affected 135,000 pieces of data held by the Police National Legal Database. The stolen information included contact details linked to parents, education staff, police officers and members of the public, although the DfE says no financial information was accessed.


Anthropic opens Claude for Teachers free to U.S. K-12 educators

Anthropic has opened a free version of Claude to verified U.S. K-12 teachers, combining premium capabilities, Cowork, teaching Skills and connectors to curriculum resources and academic standards. Educators who register by June 30, 2027 will receive a full year of access, while a separate product for schools and districts is being developed.

OpenAI recruits paid Campus Leads to run ChatGPT and Codex workshops across eight countries

OpenAI is recruiting undergraduate pairs to organize ChatGPT and Codex workshops, weekly project sessions and campus showcases across eight countries. Selected Campus Leads will receive a stipend, ChatGPT access, Codex credits, training and event funding, with applications closing on August 10.

ChatGPT for Education publishes workflow collection for faculty, students and teachers

ChatGPT for Education has published a role-based collection of workflow examples for higher education faculty, students and K-12 teachers. The resources cover course reviews, assessment calibration, research, lesson planning and student projects, with an emphasis on editable outputs, source evidence and human review.

Sal Khan says early Khanmigo fell short as Khan Academy rebuilds AI tutor

Sal Khan has acknowledged that the original version of Khanmigo did not improve student learning as much as Khan Academy had hoped. The nonprofit has now embedded the AI tutor more deeply into student practice, allowing it to respond to specific problems, question learners about their reasoning and draw on recent performance information.


Kanishka Narayan promoted to UK AI minister as Andy Burnham takes power

Kanishka Narayan has been promoted to Minister of State for Artificial Intelligence and will attend Cabinet under Prime Minister Andy Burnham. The Vale of Glamorgan MP says Britain has a narrow window to influence how AI affects employment, industry, public services and national security.

Teachers put school funding first for new UK Education Secretary Lucy Powell

School funding should be Lucy Powell’s first priority as Education Secretary, according to 72% of teachers responding to a weighted Teacher Tapp survey in England. SEND and alternative provision placed a distant second at 13%, followed by teacher recruitment at 8%.

Global survey of 45,000 finds universities losing the AI race with their own students

A global Digital Education Council survey found that 88% of students use AI in their learning, while 57% say their assessments provide inadequate guidance on its use. The findings point to widening gaps around institutional policy, faculty confidence, assessment design and students’ preparedness for AI-enabled workplaces.

Grow with Google founder Lisa Gevelber on building AI fluency in higher education

Grow with Google founder Lisa Gevelber spoke to ETIH about academic credit for AI training, changing employer expectations and the practical meaning of AI fluency. She also addressed faculty confidence, academic integrity and concerns about the influence technology companies could have over how universities define essential AI skills.

WGU and Anthropic plan AI-native learning and credentialing model

Western Governors University and Anthropic are developing an AI-native model intended to identify emerging workforce skills, personalize learning, verify capabilities through digital credentials and connect learners with career opportunities. Anthropic will provide engineering support, model credits and platform resources, while WGU deploys Claude for Enterprise across the organization.


AI gets a body as humanoid robot enters New York classroom

Realbotix is testing its Optio AI teaching assistant alongside an M-Series humanoid robot at Salamanca City Central School District in New York. The pilot begins with high school students taking AI and robotics courses, before a planned expansion to around 500 learners during the fall semester.


Will.i.am, Kahoot and ElevenLabs founders back AI-powered literacy platform imagi

AI literacy platform imagi has raised $4.5 million in a seed round led by Brighteye Ventures and Day One Capital. Its backers include Will.i.am, ElevenLabs founder Mati Staniszewski and investors connected to Kahoot, while the company says its supervised platform is already used across more than 100 U.S. school districts.


Efekta appoints former Microsoft and Amazon technology leader Matt Fisher as CTO

Matt Fisher has joined Efekta Education as Chief Technology Officer, taking responsibility for technology strategy, product development and the company’s AI capabilities. The former Microsoft and Amazon technology leader will support Efekta’s agentic teaching platform, which works with more than 4.5 million active students.

Kristen Swanson leads Anthropic Education Labs hiring push

Kristen Swanson has announced that she will lead Education Labs at Anthropic as the team recruits an education researcher and design engineer. The group is investigating how people develop lasting capability with AI, prioritizing fluency and agency over tutorials, onboarding flows and engagement metrics.


Keep up with the latest edtech news

The pace of change across AI, education policy, funding and classroom technology shows no sign of slowing. Sign up for ETIH’s daily updates to receive the latest edtech news, analysis, interviews and industry developments directly in your inbox.

Next
Next

Imperial College London plans AI, economics and policy master’s for 2027