Last updated: September 16, 2026
Since the last roundup: ETS launched an official AI-powered TOEIC Speaking app, two different parts of the US government put out conflicting signals on classroom AI, OpenAI shipped a new flagship model that underlies a lot of the prompts teachers already use, and a small new study complicates the case for digital vocabulary tools.
Industry News
US Ed Department Tells States to Judge AI Tools by Evidence, Not Screen Time. On August 20, the US Department of Education told states to evaluate classroom software by demonstrated learning gains rather than time spent on screens, in a letter from Kirsten Baesler at the Office of Elementary and Secondary Education, according to the Department’s press release. The letter names Louisiana, Arkansas, Indiana, Michigan, and Texas as states piloting outcomes-based ed-tech contracts, where vendors are paid based on results. Two weeks later, New York City took the opposite approach and banned generative AI outright for students through 8th grade. The two documents point in different directions: one asks for evidence before restricting, the other restricted first and is spending a year gathering evidence.
OpenAI Ships GPT-6 Astra, Its New Flagship Model. OpenAI released GPT-6 Astra on September 3, describing it as its most capable model yet across computer use, coding, and science, per OpenAI’s own announcement. The company said Astra is the first of its models to reach a “Critical” capability level on cybersecurity under its internal safety framework, which delayed the rollout and added extra safeguards following a security incident involving an earlier model in July. For teachers building materials with ChatGPT, including the prompts on this site’s AI prompt guide, the model behind those prompts just changed, so results may look a little different than they did last month.
Tool Updates
ETS Launches an AI-Powered TOEIC Speaking Prep App. ETS rolled out TOEIC Pal globally on July 28, an official AI-powered app for TOEIC Speaking practice, following an earlier pilot in Japan, according to ETS’s own announcement. The app covers all four TOEIC Speaking task types (Read Aloud, Describe a Picture, Respond to Questions, Express an Opinion) with official content, model answers, and instant AI scoring. TOEIC scores factor heavily into Korean hiring and graduation requirements, so this is worth knowing about alongside the material already on this site’s TOEIC and IELTS practice page.
Research
Small Study Finds a Trade-off Between Digital Tools and Focus in Vocabulary Learning. A quasi-experimental study of 87 EFL university students, published in the open-access World Journal of English Language, compared a group using interactive digital vocabulary tools against a group following a “digital minimalism” approach with fewer tech distractions, per the published study. The digital-tools group scored higher on the post-test, which the authors attribute to richer feedback. The digital-minimalism group showed better focus, fewer distractions, and stronger task engagement, even with slightly lower test scores. One caveat: this is a single-institution study with a modest sample, so it’s a data point, not a verdict on which approach wins.