< img src = "/uploads/blogs/e4/ab/ib-freeu7f19_240df4a4.jpg" Alt = "serious threat. News from Openai determines the address in the photo"/> ~ ~ ~ ~ < p > The latest artificial intelligence models from & nbsp; openai & nbsp; o3 and & nbsp; o4-mini are still & nbsp abuse.
< p dir = "ltr" >Last week, Openai released its latest artificial intelligence models, O3 and & nbsp; O4-mini. Both can & nbsp; & laquo; reason & raquo; for & nbsp; help downloaded images. Models can also cut, rotate and & nbsp; enlarge photos & mdash; even blurred and & nbsp; distorted & mdash; for their careful analysis. Users quickly realized that & nbsp; this & nbsp; can be a convenient tool for determining location on & nbsp; photo. O3, in particular, is quite good to recognize cities, prominent memory; rsquo; pits and & nbsp; even restaurants and & nbsp; bars for & nbsp; thin visual prompts.
~ ~ ~ > 62 > 62 > < p dir = "ltr" > Similar opportunities were in & nbsp; catgpt yet & nbsp; to & nbsp; nbsp; launch O3 and O4-mini, but as & nbsp; nbsp; nbp; The knife, for example, is older GPT-4O. The presence of such functionality creates potential problems for privacy. Nothing & nbsp; hinders to malefactors, a shot, to make a screenshot instagram and & nbsp; to determine the location thanks to chatbot to pursue the victim.
< p dir = "ltr" > There are now little precautionary measures to prevent this type of application Chatgpt. And Openai not & nbsp; see in & nbsp; this problem.
< p dir = "ltr" > & laquo; openai o3 and o4-mini add visual arguments in & nbsp; chatgpt, making it more useful in & nbsp; on & nbsp; emergency situations. We & nbsp; have worked to teach our models to reject requests to & nbsp; nbsp; confidential or confidential information, added protection tools aimed at & nbsp; nbsp; to prohibit models to identify individuals on & nbsp; nbsp; nbsp; nbsp; nbsp; nbsp; nbsp; nbsp; nbsp; nbsp; says company representative in & nbsp; comments Techcrunch.