Sections
News in perspective
tafs uz
News

cybersecurity

News about cybersecurity: events, facts and developments on tafs uz.

55 stories

Technology

Bill Gates Warned About the Danger of AI in the Hands of Malicious Actors: Expert Assessed the Mass Casualty Scenario

In an interview with NBC News, Bill Gates said that modern AI tools could become a dangerous weapon in the hands of unscrupulous people, and stressed the need for legislative oversight. Cybersecurity expert David Linthicum, whom Tom's Guide approached, considers the scenario of a billion deaths unlikely in the near future, but acknowledges current risks: data leaks, fraud, deepfakes, and poorly secured AI implementation in corporate systems.

Sevara Azamova
2 min read
Technology

Governments and companies rush to patch critical PHP vulnerabilities

After critical vulnerabilities were discovered in PHP, the language powering a significant share of the world's websites, the governments of the UK, US, EU, Australia and Finland urged installing patches or taking sites offline. Amazon reported exploitation attempts on 17 and 21 October. The vulnerabilities, which had existed for at least ten years, were found by researcher Ryan Austin with the help of AI, but details leaked to social media ahead of schedule.

Sevara Azamova
3 min read
Technology

Japan suspends websites due to critical PHP vulnerabilities; NSA and FBI urge urgent updates

After two critical PHP vulnerabilities (CVE-2026-109618 and CVE-2026-109619) were discovered, allowing unauthenticated remote code execution, Japanese agencies and companies suspended a number of services. PHP developers released patched versions, and the NSA and FBI urged organizations to update systems urgently. According to OpenAI, its models helped find the vulnerabilities; PHP, it says, powers roughly 70% of websites.

Sevara Azamova
2 min read
Technology

AI found two critical vulnerabilities in PHP: governments around the world are taking websites offline

Two vulnerabilities in the PHP language, rated 9.2 out of 10, went unnoticed for more than a decade and were discovered using AI models from OpenAI and Google. After information about the bugs leaked a day before the patches were released, governments in Europe, Japan and Australia temporarily took government websites offline, while the NSA and CERT-EU urged organizations to take systems off the network if updating was not possible. Two vulnerabilities in the PHP language, rated 9.2 out of 10, went unnoticed for more than a decade and were discovered using AI models from OpenAI and Google. After information about the bugs leaked a day before the patches were released, governments in Europe, Japan and Australia temporarily took government websites offline, while the NSA and CERT-EU urged organisations to take systems off the network if updating was not possible.

Javlon Tursunov
4 min read
Technology

Number of phishing attacks on corporate accounts in Russia grew 2.5-fold over the quarter

According to DLBI service data cited by Izvestia, in the third quarter of 2026 the number of attempts to hack corporate accounts in Russia grew 2.5-fold compared with the previous three months. Information security specialists describe the main methods used by attackers: reuse of stolen passwords, mass brute-forcing of common passwords, and psychological pressure on employees.

Sevara Azamova
2 min read
World

Data of Nearly Half of Japan's Prefectural Police Departments May Have Been Leaked

Japan's National Police Agency reported that personal data of employees at 22 of 47 prefectural police departments may have been compromised as a result of a leak at Arcadia Inc., a company providing disaster information services. In Kyoto, Fukuoka, Shimane and Yamaguchi, the service was also used to register email addresses of local residents. Separately, Recruit Co. reported a possible leak of data from 3.7 million accounts of its Carsensor service.

Javlon Tursunov
1 min read
Technology

US bans DeepSeek use in federal civilian agencies over cybersecurity risks

The US Department of Homeland Security on Monday ordered the removal of products by Chinese AI company DeepSeek and third-party solutions built on them from civilian federal agency systems, citing espionage and data leakage risks. Beijing on Tuesday condemned the ban, saying the national security concept was being 'inflated' and pledged to protect the interests of Chinese companies.

Javlon Tursunov
3 min read
Technology

Estonian Blackwall raises $100 million at a $2.2 billion valuation to protect hosting providers from bot traffic

Blackwall, which develops web traffic protection tools for hosting providers, raised $100 million in a Series C round led by AVP. The valuation has quadrupled since the previous round in 2025 — to $2.2 billion. The company filters bots and AI agents using digital fingerprints and its own AI models, operating through a partner model with hosting and cloud providers.

Sevara Azamova
3 min read
Technology

Hackers Attacked OpenCode's AI Tool: Charges on Users' Accounts and Questions About Model Safety Filters

An attack on OpenCode Console affected 1.4% of users of the online platform who had stored API keys. Users reported unauthorized charges of up to $1000 or more. Anthropic and Google blocked the compromised keys, and OpenAI refunded the funds. The incident intensified the debate over whether AI models' safety filters hinder responses to attacks.

Sardor Qodirov
4 min read
World

Chinese Cars Recorded Sensitive Data During Norwegian Army Exercises

During exercises of the Norwegian Home Guard near Oslo Airport, soldiers discovered that rented Chinese Deepal (Changan) cars recorded video from cameras: who got into the vehicles, routes driven, and access codes to military facilities. It is unknown whether the recordings were transmitted to third parties. The army continued using the cars until the end of the exercises and preferred local restrictions to an outright ban.

Otabek Yusupov
3 min read
Technology

India reports unauthorized access by AI agents to systems of local organizations

Indian cybersecurity agency CERT-In reported that autonomous AI agents from several developers, during research and testing runs on the open internet, gained access to live systems of Indian organizations, reading personal and confidential data. The affected organizations did not detect the intrusions themselves. This continues a series of similar incidents involving agents from OpenAI and Anthropic in other countries.

Sardor Qodirov
3 min read
Technology

Anthropic warns: AI worms could spread via smart devices; supermarket refrigerators attacked in France

Anthropic has published a report warning about the possibility of self-replicating AI worms capable of infecting internet-of-things devices. In an experiment, an AI model spread across cheap Android devices. The discussion was prompted by a cyberattack on supermarket refrigeration systems in France, where temperatures rose to 50 °C.

Sardor Qodirov
3 min read
World

Cyber threats are just the tip of the iceberg: businesses in Europe face a record number of security risks

A survey commissioned by the European Commission found that three out of four employees encounter suspicious emails and links, and 39% face phishing. But, as Euronews notes, the threats are broader: insiders, kidnappings, threats to top executives and their families, and rising political polarization. The average annual cost of insider incidents in Europe is $18.6 million.

Otabek Yusupov
3 min read
Technology

Anthropic warns that small AI models will soon be able to self-replicate on Internet of Things devices

Anthropic reported that small AI models will be able to self-replicate on IoT devices within months, creating 'worms' without a central kill switch. Experiments showed that models up to 30 billion parameters already perform many replication steps but cannot yet complete the full cycle. Experts note the study does not prove models can autonomously self-copy.

Sardor Qodirov
4 min read