Anthropic says GPT 5.5 also has similar issues, directly dragging OpenAI into the water, Silicon Valley infighting scene +1

CryptoWorld News reports that, according to The Wall Street Journal, a research study submitted by Amazon indicates that Fable 5 can be induced to output sensitive information. AWS CEO Andy Jassy will discover and report this to the U.S. government, prompting the White House to restrict foreign users’ access to the Anthropic model. Anthropic denies that this issue constitutes jailbreaking, stating that similar vulnerabilities also exist in models such as GPT 5.5. Former U.S. Commerce Department official Kate Colen said that disagreements between Anthropic and the White House on AI safety boundaries have affected decision-making.