WaperOpenAI is reportedly nearing a funding round of up to $30 billion after lowering its valuation target.
OpenAI tracker5 items
Security also dominated the news, with new offline-training safeguards and an external audit finding substantial gaps in agent monitoring.
New4
OpenAI reportedly lowers valuation target for funding round of up to $30 billion
The Decoder reports that OpenAI is nearing a new funding round of up to $30 billion after lowering its valuation target. The round has not been reported as closed.
Why it matters: The prospective raise would strengthen OpenAI’s funding, while the reduced target signals pressure on its valuation expectations.
OpenAI strengthens frontier training containment with offline environments
OpenAI detailed stronger containment for frontier-model training and evaluation, including offline environments and high-compute red-teaming. Bloomberg reports that the move to training without internet access follows cyber incidents.
Why it matters: The changes address security risks in the environments used to develop and evaluate OpenAI’s most capable models.
Texas sues OpenAI over alleged deceptive practices and child safety
Texas’s attorney general sued OpenAI over alleged deceptive practices and child-user safety issues involving its chatbot, according to reports from Yahoo, The Texas Tribune and Bloomberg Law.
Why it matters: The lawsuit adds legal exposure around ChatGPT’s consumer claims and protections for younger users.
OpenAI details storage changes delivering 12× faster ChatGPT reads
OpenAI says it scaled its Habitat storage system to serve more than one billion ChatGPT users. The architecture uses multi-tenancy, global caching and Azure Cosmos DB to deliver 12× faster reads.
Why it matters: The technical update shows how OpenAI is adapting ChatGPT’s infrastructure to support its stated user scale.
Follow-ups1
External audit finds substantial gaps in OpenAI agent monitoring
METR and Redwood Research spent three weeks auditing OpenAI and found up to 25% of agent traffic from GPT-5.5-class models and above went unmonitored. OpenAI redacted large portions of the 61-page report; only a summary is public.
Why it matters: The findings raise questions about how reliably OpenAI detects unsafe agent behavior as customers delegate more consequential tasks.