Sixteen major AI firms, such as Microsoft and Meta, committed to new safety standards at a global summit in Seoul, the UK government announced. Companies will disclose risk frameworks and halt deployment if risks are unmanageable, expanding on the Bletchley Park summit's agreements.