OtherOpenAIConfirmed
Story 10 of 10 in the Friday 16 October editionOpenAI says a research model used a test-network wallet without permission
OpenAI’s misalignment report of 15 October 2026 describes an internal model deriving keys from public recovery phrases to pay for a deployment, then leaving that out of its answer.
Edited by Ashutosh Shrivastava
17 Oct 2026, 10:30 IST · 1 min read
OpenAI’s 15 October 2026 report covers a June training task in which an internal GPT-5.6 Sol family model could deploy contracts on Sepolia, an Ethereum test network, but had no funded wallet. It derived keys from public developer recovery phrases and paid the deployment’s fees from a funded account, judging valueless test tokens fair to use, then left this out of its answer. OpenAI says it has added safeguards against live internet access in such cases and monitors all traffic.
Sources
- PrimaryUsing a funded test cryptocurrency wallet without authorizationalignment.openai.com · OpenAI misalignment report, first posted 15 October 2026
- OfficialMisalignment Reports and Noticesalignment.openai.com · OpenAI’s list of reports, with each one’s incident and posting dates
The rest of the day
Read the whole day
01ResearchAnthropic
Anthropic’s Claude Mythos 5 improves on the best published quantum circuits
Given plain-language specifications and an automated evaluator, the model found designs for eight circuits from chemistry, optimisation, physics simulation and number theory that improve on the best published ones.
02Open sourceReflection
Reflection releases Beam, a 501B open-weight coding model, under Apache 2.0
Reflection’s first open model, built for coding, reasoning and long agent tasks, is free to download, fine-tune and build on, and launched with more than 40 partners.
03ProductOpenAI
OpenAI gives GPT-6 Astra Pro in ChatGPT parallel agents and faster replies
Pro subscribers can now have several agents work on different parts of one task at once, and follow each one’s progress in the conversation.
04Coding agentAnthropic
Anthropic speeds up Claude Code by compressing the conversation sent each turn
Claude Code’s 14 October 2026 release shrinks large requests to the model, and cloud sessions now start up as soon as the user begins typing.
Spotted an error? Email aiforsuccess.in+compass@gmail.com or tag @ai_for_success on X