"According to United Nations figures, a child is abused every 30 minutes in Congo." (translated)
Des agents incontrôlés d'OpenAI ont effacé une partie de leurs traces
www.lalibre.be
"act in the right way." Confronted with increasingly vivid concerns about the development of artificial intelligence, Donald Trump explained on Tuesday that self-regulation remains his preferred approach and assured that the big bosses of the sector, whom he received, wanted "to act in the right way." (translated)
Des agents incontrôlés d'OpenAI ont effacé une partie de leurs traces
www.lalibre.be
Agents of artificial intelligence from OpenAI, acting outside their framework, attempted to erase traces of their access to government sites, according to a report by experts, without being able to determine if this action was deliberate. (translated)
Des agents incontrôlés d'OpenAI ont effacé une partie de leurs traces
President Donald Trump has stated in an interview that the U.S. government could consider buying shares in artificial intelligence companies such as OpenAI and Anthropic, similar to what it has already done with Intel. (translated)
Trump asegura que EEUU puede comprar acciones de OpenAI y Anthropic
The American Federal Trade Commission has launched an investigation against OpenAI and Anthropic to assess the adequacy of security measures in relation to the risks posed by advanced artificial intelligence. (translated)
La Commission fédérale américaine du commerce lance une enquête contre OpenAI et Anthropic pour prévenir les risques liés aux IA de pointe
The US consumer protection agency FTC has launched a comprehensive investigation into the security risks of AI companies like OpenAI and Anthropic, as the public debate about potential threats from AI systems intensifies. (translated)
Künstliche Intelligenz: US-Behörde nimmt KI-Firmen stärker in den Fokus
Google has unveiled a new AI model Gemini 4, named Argon, which is considered its most powerful product to date, comparable to the top models from companies OpenAI and Anthropic, although its public release remains unknown. (translated)
Po měsících odkladů byl představen AI model Gemini 4
Premier Australii Anthony Albanese poinformował 23 września, że agenci AI firmy OpenAI włamali się do rządowego systemu statystycznego, naruszając "niewrażliwe" dane tamtejszego systemu ochrony zdrowia. #AI #OpenAI #cyberbezpieczeństwo
AI próbowała włamać się na stronę rządu Kanady. To znowu OpenAI?
wiadomosci.wp.pl
Autonomous artificial intelligence agents made an unsuccessful attempt to break into the Canadian government's website; however, the Canadian Cyber Security Centre confirmed that no government systems were compromised. (translated)
AI próbowała włamać się na stronę rządu Kanady. To znowu OpenAI?
The Federal Trade Commission (FTC) has launched a comprehensive investigation into AI risks at the companies Anthropic and OpenAI, following reports of security incidents and malfunctioning AI systems. (translated)
US-Untersuchungen bei Anthropic und OpenAI
A
Președintele FTC, Andrew Ferguson, avertizase deja că dezvoltatorii care folosesc agenți AI în teste de securitate și ajung să provoace atacuri ar putea fi considerați responsabili pentru eventualele prejudicii. #AI #OpenAI #FTC
SUA deschid o investigație majoră asupra OpenAI și Anthropic. Ce riscuri prezintă agenții AI pentru utilizatori
www.mediafax.ro
American authorities have opened a major investigation into OpenAI and Anthropic to analyze the risks that AI agents present to users, considering recent incidents in which they demonstrated the ability to exploit vulnerabilities in external systems. (translated)
SUA deschid o investigație majoră asupra OpenAI și Anthropic. Ce riscuri prezintă agenții AI pentru utilizatori
D
"Ako AI sustavi jednog dana dobiju moć i više nas ne budu morali slušati, svijet će početi oblikovati prema vlastitim ciljevima. A to za nas vjerojatno neće biti dobro""Ako AI sustavi jednog dana dobiju moć i više nas ne budu morali slušati, svijet će početi oblikovati prema vlastitim ciljevima. A to za nas vjerojatno neće biti dobro", smatra Daniel Kokotajlo, bivši istraživač OpenAI-ja. #UmjetnaInteligencija #TehnološkiRazvoj #EtikaAI
AI probio granice koje nije smio: Upozorenja sada stižu i iz kompanija koje ga razvijaju
dnevnik.hr
"AI model koji je trenirao OpenAI pristupio je četirima javnim internetskim stranicama u Australiji""AI model koji je trenirao OpenAI pristupio je četirima javnim internetskim stranicama u Australiji", kazao je Richard Marles, potpredsjednik australske valde. #UmjetnaInteligencija #TehnološkiRazvoj #EtikaAI
AI probio granice koje nije smio: Upozorenja sada stižu i iz kompanija koje ga razvijaju
dnevnik.hr
In the podcast 'It's Not That Simple', Marike Stellinga and Maarten Schinkel investigate the risks of circular financing between large AI companies, chipmakers, and data center builders, and they raise questions about the valuations of companies like OpenAI and Anthropic, which are estimated to be very high despite a lack of profit. (translated)
Worden de waardes van AI-bedrijven kunstmatig opgeblazen?
L
Google launched its new artificial intelligence model, Gemini 4, on Wednesday, but is limiting access to a select group of cybersecurity experts through the Fairwind program, following in the footsteps of its competitors Anthropic and OpenAI. (translated)
Google lance un nouveau modèle d'IA mais en limite l'accès à des experts en cybersécurité
In the Tech Flash of September 30, Léa Benaim discusses OpenAI's $30 billion fundraising, as well as other topics related to AI and technology. (translated)
Le Tech Flash : OpenAI veut lever 30 milliards de dollars, par Léa Benaim - 30/09
An investigation has been launched in the U.S. against AI companies Anthropic and OpenAI due to the potential risks of their technologies for consumers, following a series of incidents that have raised public concern. (translated)
U SAD-u kreće istraga protiv AI divova
OpenAI has decided to halt the release of its new model GPT-6.1 Astra, as the model showed serious errors, including deception and decision-making without the user's knowledge. (translated)
Uus OpenAI supermudel pannakse pausile – te aimete ilmselt juba põhjust
J
"Tas, kas šeit ir zīmīgi, ka "OpenAI" savos testos modelim lika iegūt datus. Un, kad tas sastapās ar šķēršļiem un nevarēja iegūt nepieciešamos datus, tas nolēma mēģināt izmantot ievainojamības. Mēs neredzējām ļoti dziļu ielaušanos, taču tas ir brīdinājuma signāls: mākslīgā intelekta aģenti, cenšoties sasniegt bieži vien pavisam ikdienišķus mērķus, var ķerties pie sistēmu ievainojamību izmantošanas."Kāpēc aģents, pildot šķietami vienkāršu uzdevumu, vispār ķērās pie sistēmas ievainojamībām, "Pasaules panorāmai" skaidroja kiberdrošības pētnieks Džeks Keibls: #MākslīgaisIntelekts #Kiberdrošība #Progresība
Mākslīgais intelekts: kad cīņa par progresu nomāc brīdinājumus par draudiem
www.lsm.lv
