Sam Altman, CEO of OpenAI, says testing GPT-5 left him nervous, comparing its development to the Manhattan Project. As the new AI model nears an August launch, Altman has raised urgent concerns about its capabilities and the absence of proper oversight. At the same time, fraud experts say generative AI is already being weaponised at scale. With pressure mounting from investors and Microsoft, GPT-5 may not just be a technical leap but a flashpoint for AIs future direction.