IT News
OSN chce překreslit mapu světa. Co to vlastně znamená v praxi, proč nám mapy lžou a jak vypadá ideální Equal Earth
Srpnové aktualizace Windows 11 se nevydařily. Ze záhrobí se vrací ukazatel z Windows XP
„Rozumní zákazníci vědí, že při nákupu hry platí za licenci,“ říká Sony k soudnímu sporu o vlastnictví
V Havířově mají chytré křižovatky. V noci svítí na semaforech červená, když přijede auto, dají mu zelenou
Proč trvají aktualizace na nový Android tak dlouho – neukazujte prstem jen na výrobce, ale i na Google
Díky novým PowerToys se snáze zorientujete v záplavě oken a jas změníte na více obrazovkách najednou
NASA vypustila nový vesmírný dalekohled. Oblohu dokáže mapovat 1000krát rychleji než Hubbleův teleskop
Raiffeisenbank vrací 50 % a Fotoškoda dá slevu 25 000 Kč na Nikon
OpenAI agents discussed ways to escape their sandbox on public wiki
Self-identifying OpenAI agents posted 18,000 messages to a public wiki that discussed ways for other agents to bypass security sandbox restrictions during what was likely internal testing designed to gauge the agents’ hacking abilities, researchers said Friday.
In all, agents with 3,700 distinct self-given names posted the messages to German site DSEwiki over a six-week period. Besides discussing ways the agents could break out of the restricted environment OpenAI intended to prevent them from posting code or content to the Internet, the posts shared test answers. The posts also shared possible ways to perform XSS (cross-site scripting) attacks against the wiki and to impersonate site moderators. In three of the posts, agents used the word “swarm” to describe the collection of agents engaged in the activity.
Colluding to share answersThe research team—composed of Sydney Von Arx, Spencer Kitts, Thomas Larsen, and Cormac Slade Byrd—said they found the posts and pieced them together. The researchers say there are gaps in their understanding of precisely what actions the agents took because the research is based solely on the content of the posts. Additionally, the agents generated “chain of thought” data that’s understood only by OpenAI. As a result, the researchers said, they in some cases made educated guesses, including that the agents were, in fact, from OpenAI. In a statement, OpenAI later confirmed they were.
Xiaomi přivezlo do Evropy nového robota CyberOne. Rozhoduje se sám a sbírá zkušenosti v továrně na elektroauta
“Trust, not features, is the real deficit”: VMware tries to appease SMBs
For many small-to-medium-sized businesses (SMBs), VMware has become too expensive.
Broadcom’s acquisition of the virtualization firm brought the end of perpetual license sales and the arrival of pricey, stacked, subscription-based bundles that priced out many SMBs.
The most obvious is VMware Cloud Foundation (VCF), VMware’s flagship private cloud bundle that has been Broadcom’s primary focus since taking over VMware. Many SMBs find that VCF is unaffordable and stuffed with unnecessary offerings. However, numerous customers have reported online that VMware sales representatives have still pushed them toward VCF, with some claiming that sales reps have told them that the lower-priced edition of VMware’s virtualization platform, vSphere Standard, was no longer available.
Once popular for attacking AI, ASCII smuggling is embraced by spammers
A clever technique used to hide malicious prompts in attacks on AI agents has been adopted by spammers to evade filters on email platforms that are designed to flag unwanted messages used in mass campaigns.
The technique is broadly known as ASCII smuggling. It gained attention two years ago as a means of making a class of AI attack known as prompt injections more stealthy. Malicious instructions embedded in emails or other untrusted content to be processed by an LLM aren’t written in ordinary text. Instead, they’re rendered by a special range of Unicode tags. For example, the tag point U+E0041 mirrors “A,” and U+E0061 mirrors “a.”
No longer just for obscuring prompt injectionsThe block of 128 tags mimics a portion of the American Standard Code for Information Interchange almost perfectly, with one major difference: the characters they encode are readable by computers but, by design, are almost completely invisible to humans. By expressing the malicious prompts in these tags, LLMs detect the instructions, but people reading the email never see them. There’s much more about ASCII smuggling here.
Za nabídkou Start ve Windows 95 stojí psycholog, který původně zkoumal komunikaci šimpanzů
Netflix a další na víkend: Chad Powers a nová sezóna Gentlemanů. Mayday nebo Star Wars: Mandalorian a Grogu
Všichni o něm jen mluví, ale v Berlíně si už na něj sáhnete. Drželi jsme v ruce skládací iPhone
Zadání znělo: Najdi na webu nějaká čísla. Roj agentů OpenAI si k tomu zřídil další tajnou nástěnku
Německo odpálilo balistickou raketu. Naposledy je mělo ve výzbroji před více než třiceti lety
Vše, co Apple ukáže příští týden na keynote: tři iPhony, dvoje Apple Watch a nová sluchátka
Cyberpunkové město z ASCII znaků působí skoro jako skutečný svět. Nechybí auta, chodci ani budovy s interiéry
Konzolová legenda se promění v Lego stavebnici. Útroby PlayStationu skrývají odkazy na herní klasiky
- « první
- ‹ předchozí
- …
- 9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17
- …
- následující ›
- poslední »



