Toggle light / dark theme

Chimpanzees rationally revise their beliefs

The study, titled “,” was conducted by a large research team that included UC Berkeley Psychology Postdoctoral Researcher Emily Sanford, UC Berkeley Psychology Professor Jan Engelmann and Utrecht University Psychology Professor Hanna Schleihauf. Their findings showed that chimpanzees — like humans — can change their minds based on the strength of available evidence, a key feature of rational thought.

Working at the Ngamba Island Chimpanzee Sanctuary in Uganda, the researchers presented chimps with two boxes, one containing food. Initially, the animals received a clue suggesting which box held the reward. Later, they were given stronger evidence pointing to the other box. The chimps frequently switched their choices in response to the new clues.

“Chimpanzees were able to revise their beliefs when better evidence became available,” said Sanford, who is a researcher in the UC Berkeley Social Origins Lab. “This kind of flexible reasoning is something we often associate with 4-year-old children. It was exciting to show that chimps can do this too.”

To ensure the findings reflected genuine reasoning rather than instinct, the team incorporated tightly controlled experiments and computational modeling. These analyses ruled out simpler explanations, such as the chimps favoring the latest signal (recency bias) or reacting to the most obvious cue. The models confirmed that the chimps’ decision-making aligned with rational strategies of belief revision.

“We recorded their first choice, then their second, and compared whether they revised their beliefs,” Sanford said. “We also used computational models to test how their choices matched up with various reasoning strategies.”

The study challenges the traditional view that rationality — the ability to form and revise beliefs based on evidence — is exclusive to humans.

“The difference between humans and chimpanzees isn’t a categorical leap. It’s more like a continuum,” Sanford said.

As AI grows smarter, it may also become increasingly selfish

New research from Carnegie Mellon University’s School of Computer Science shows that the smarter the artificial intelligence system, the more selfish it will act.

Researchers in the Human-Computer Interaction Institute (HCII) found that (LLMs) that can reason possess selfish tendencies, do not cooperate well with others and can be a negative influence on a group. In other words, the stronger an LLM’s reasoning skills, the less it cooperates.

As humans use AI to resolve disputes between friends, provide marital guidance and answer other social questions, models that can reason might provide guidance that promotes self-seeking behavior.

Glowing Green: A Quantitative Analysis of Photoluminescence in Six North American Bat Species

WhatsApp is rolling out passkey-encrypted backups for iOS and Android devices, enabling users to encrypt their chat history using their fingerprint, face, or a screen lock code.

Passkeys are a passwordless authentication method that allows users to sign in using biometrics (such as face recognition or fingerprint), PINs, or security patterns instead of traditional passwords. They enable logging into websites, online services, or apps without needing to remember complex passwords or use a password manager.

When creating a passkey, your device generates a unique cryptographic key pair consisting of a private key stored on the device and a public key sent to the website or app. Because of this, passkeys provide significantly improved security over regular credentials, seeing that they can’t be stolen in data breaches because the private key never leaves your device.

Massive surge of NFC relay malware steals Europeans’ credit cards

Near-Field Communication (NFC) relay malware has grown massively popular in Eastern Europe, with researchers discovering over 760 malicious Android apps using the technique to steal people’s payment card information in the past few months.

Contrary to the traditional banking trojans that use overlays to steal banking credentials or remote access tools to perform fraudulent transactions, NFC malware abuses Android’s Host Card Emulation (HCE) to emulate or steal contactless credit card and payment data.

They capture EMV fields, respond to APDU commands from a POS terminal with attacker-controlled replies, or forward terminal requests to a remote server, which crafts the proper APDU responses to enable payments at the terminal without the physical cardholder present.

Introducing Aardvark: OpenAI’s agentic security researcher

Aardvark represents a breakthrough in AI and security research: an autonomous agent that can help developers and security teams discover and fix security vulnerabilities at scale. Aardvark is now available in private beta to validate and refine its capabilities in the field.

Aardvark continuously analyzes source code repositories to identify vulnerabilities, assess exploitability, prioritize severity, and propose targeted patches.

Aardvark works by monitoring commits and changes to codebases, identifying vulnerabilities, how they might be exploited, and proposing fixes. Aardvark does not rely on traditional program analysis techniques like fuzzing or software composition analysis. Instead, it uses LLM-powered reasoning and tool-use to understand code behavior and identify vulnerabilities. Aardvark looks for bugs as a human security researcher might: by reading code, analyzing it, writing and running tests, using tools, and more.

AI efficiency advances with spintronic memory chip that combines storage and processing

To make accurate predictions and reliably complete desired tasks, most artificial intelligence (AI) systems need to rapidly analyze large amounts of data. This currently entails the transfer of data between processing and memory units, which are separate in existing electronic devices.

Over the past few years, many engineers have been trying to develop new hardware that could run AI algorithms more efficiently, known as compute-in-memory (CIM) systems. CIM systems are electronic components that can both perform computations and store information, typically serving both as processors and non-volatile memories. Non-volatile essentially means that they can retain data even when they are turned off.

Most previously introduced CIM designs rely on analog computing approaches, which allow devices to perform calculations leveraging electrical current. Despite their good energy efficiency, analog computing techniques are known to be significantly less precise than digital computing methods and often fail to reliably handle large AI models or vast amounts of data.

/* */