Gamepad — Gaming market, technology, information security, and politics

All Cyber security Financial technology Gaming market Politics Technology

Security researchers have uncovered a massive exposure: more than 35,000 apps built with Grok are reachable online without a sign-in, and some are spilling confidential business records, employee lists, and even plain-text passwords. The findings turn a flashy AI app builder into a serious data privacy wake-up call.

Most of the exposed apps were games, landing pages, and calculators, but roughly one in ten contained internal or personal data, including staff apps from X and SpaceXSI. Researchers are now pressing for safer defaults and domain-level access controls, while affected organizations face a race to lock down data that should never have been public.

Over 35,000 Grok-built apps exposed online, leaking passwords and confidential data

#cybersecurity #privacy #data-breach #grok #ai-apps #exposed-data #app-security

View detail →

NEAR AI just confirmed a major security incident: an attacker exploited a remote code execution flaw in SGLang and broke into one of its inference servers. The breach reached requests processed in confidential mode, exposing the fragile reality behind trusted execution environments when the software running inside is vulnerable.

The attacker also accessed an internal system behind the deprecated Private Chat product, downloading names and emails for about 100 users plus plan and balance data for some accounts. On top of that, 18 AI agents on an older deployment were compromised, with chat history, settings, and logs stolen, and two API keys exposed in logs.

NEAR AI says it revoked sessions and admin credentials, rotated tokens, patched the SGLang flaw, and restored access. The lesson is brutal but clear: confidential AI needs serious security engineering, not just a TEE label.

SGLang RCE attack hits NEAR AI: confidential chats, agents, and user data exposed

#ai #cybersecurity #rce #data-breach #cloud-security #near-ai #sglang

View detail →

A critical authentication bypass in Xen Orchestra is being actively exploited in the wild, and the entry fee is a single crafted cookie. Vates VMS tracks the flaw as VSA-2026-062, and there is still no CVE assigned, which means defenders are hunting a live threat with an incomplete map.

The mechanics are brutal in their simplicity: REST cookie parsing turns an authenticationToken value into an empty object, the token provider accepts it without requiring a string, and the collection returns the first matching token. From there an attacker authenticates as that token's owner and pulls reusable login token IDs from the users endpoint, gaining admin privileges, arbitrary code execution on managed pools, and full root takeover of XCP-ng nodes.

Researchers have linked the activity to infrastructure previously flagged as a RedLine Stealer command-and-control server, including a known IP and a TOR domain. Fixes are available in Xen Orchestra 6.9.1 and 6.10.0 along with xo-server 5.211.2 and 5.213.1. If patching today is not possible, keep instances off the public internet, block the known indicators, reject empty tokens at the WAF or reverse proxy, and rotate every password and secret on the box.

Xen Orchestra auth bypass is under active attack and hands out admin access for a single forged cookie

#cybersecurity #privacy #threat-intelligence #authentication-bypass #data-breach #patching #xen-orchestra

View detail →

New guidance for security leaders puts leaked credentials at the center of board-level risk. The message is blunt: accountability should not turn into blame, or exposed secrets will go underground while the danger remains live.

From measuring time-to-revocation to testing outside reporting channels, the playbook urges CEOs to ask harder questions about how fast exposed keys are found, reported, and shut down. With AI tools and agents expanding the leak surface, visibility and speed are becoming the real proof of security progress.

Leaked credentials are still a boardroom blind spot: what security leaders must tell the CEO

#cybersecurity #data-breach #ai-security #ceo #leaked-credentials #security-leadership #boardroom-risk

View detail →

A single attacker armed with a swarm of AI agents tore through emergency services, banks, universities and military systems across at least 11 countries, and one of the deepest wounds landed in Taiwan. Researchers at Acronis traced the operation to one person based in Hualien City, who ran a setup called BOSSAI with twelve agents, nine of them active, pulling dispatch records covering seven Taipei districts straight out of the 119 emergency API.

The details are wild: 90,000 files left behind on the attacker's own server, including agent conversations, command history, credentials and stolen victim data. The agents also stumbled onto a Taipei fire department platform leaking patient records from 2018, complete with ECG readings and ambulance footage, with no login required. When commercial cloud models refused to help with the attacks, the operator fell back on abliterated local models of DeepSeek, GLM and Qwen to keep the campaign running.

The bigger story is not one criminal, it is the collapse of the barrier to entry. An operation that once demanded a team and years of tradecraft now scales with rented compute and stripped-down models, while scanning tools invent vulnerabilities and report false positives. If AI agents are now hitting fire departments and disaster response centers, your own login is part of the attack surface, and encrypted connections, hardened credentials and a serious VPN are no longer optional extras.

One hacker, twelve AI agents: Taipei emergency dispatch records stolen in a 12-day rampage

#ai #cybersec #hacking #taiwan #privacy #data-breach #emergency-services

View detail →

The Italian foreign minister says cybercriminals have been publishing his personal data on the dark web for days and pressuring his family in an attempt to extract more information. He refused to be intimidated and linked the attack to Italy's support for Ukraine against Moscow.

Prime Minister Giorgia Meloni called the sale of confidential government data an issue of extreme concern and vowed to respond with maximum determination. The incident adds to a wave of disruptive cyberattacks across Europe, from manipulated supermarket refrigeration systems to false bomb threats against commercial flights.

With state-linked and criminal hacking campaigns escalating, the case puts a spotlight on personal data exposure, geopolitical retaliation and the urgent need for stronger privacy and security practices.

Italy's foreign minister targeted by pro-Russia hackers as personal data leaks on dark web

#hacking #cybersecurity #privacy #data-breach #russia #italy

View detail →

Japan is moving fast to harden its digital defenses after a series of unauthorized access incidents and data leaks hit companies and government agencies. Prime Minister Sanae Takaichi announced an emergency cybersecurity package worth about US$63 million, with a clear message: public institutions and businesses must be ready for threats amplified by artificial intelligence.

The plan includes faster public-private information sharing, rapid support for victims, and a dedicated cyber rescue team to help affected organizations recover. Japan also plans to expand its legally authorized active cyber defense capability, allowing preventive access to hacker servers when national security is at risk, while a new AI-powered expert team will hunt for vulnerabilities and track threats.

Japan launches emergency cybersecurity package after wave of attacks

#cybersecurity #privacy #data-breach #ai-threats #japan #cyber-defense

View detail →

The Fundo Garantidor de Créditos (FGC) has notified investors after detecting unauthorized access to personal data in late October. Exposed information includes names, CPF numbers, emails, birth dates, and details tied to guarantee payments, although the entity says there is no evidence so far of misuse or financial loss.

The case was reported to Brazil’s National Data Protection Authority (ANPD), and investigations continue with specialized support. Because leaked data can fuel phishing and financial scams, investors should treat suspicious messages, calls, and emails with extra caution, and reinforce their digital privacy defenses.

For anyone worried about digital exposure, strengthening privacy habits and using trusted protection like ProtonVPN is a smart move.

FGC data breach exposes investor details and triggers privacy alert

#brazil #cybersecurity #anpd #privacy #data-breach #fgc #investor-alert

View detail →

Chainstack has disclosed a serious production systems breach after an external attacker slipped in through a third-party component and escalated from a development environment into parts of its production infrastructure. The company says customer account data was accessed, including organization names, account holder names, member email addresses, and password hashes.

Chainstack also warns that systems holding node credentials were reachable, and while there is no evidence they were read, customers are being told to treat them as exposed. The incident is being linked to a broader AI-powered attack campaign tied to the ARTEX autonomous agent framework, which has also hit WorkOS, Chainalysis, and DIVD. Rotate node credentials and platform API keys, update reused passwords, and stay alert for phishing.

For security teams, this is a reminder to harden third-party access, monitor cross-system anomalies, and use encrypted connectivity tools such as ProtonVPN as part of a layered defense.

Chainstack breach exposes customer data and password hashes in AI-powered attack wave

#cyberattack #data-breach #chainstack #password-hashes #artex #blockchain-security

View detail →

Binance has disclosed a staggering breach: hackers allegedly used a fake job interview to infiltrate internal systems and steal US$ 6.3 billion in crypto assets. The attack reportedly began with a tampered AI coding tool and escalated through stolen credentials, a virtual meeting with the CEO of a fictitious company, and a malicious script executed on October 29.

The FBI has linked the operation to a North Korean hacking group, an accusation Pyongyang rejects. Binance says it is reimbursing affected customers and expanding AI-powered detection, but the incident raises urgent questions about security, accountability, and trust in centralized crypto platforms.

Fake job interview led to $6.3 billion Binance hack, FBI says

#fintech #hacking #cybersecurity #data-breach #crypto #north-korea #binance

View detail →

The numbers are staggering: 6.43 billion dollars drained from the world's largest cryptocurrency exchange in a multi-stage campaign that reads like a thriller. Binance has now published a detailed post-mortem of the attack, and the playbook it exposes is a masterclass in modern cyber warfare.

It began with something as ordinary as a fake job interview. A persona posing as a project director lured an engineer onto a chat app and handed over an assignment packed with malicious code, while another engineer was compromised earlier through a trojanized Homebrew installer for a popular AI tool. Poisoned developer tooling, it turns out, is one of the fastest ways into a billion-dollar platform.

From there the attackers abused internal AI tooling, stole deployment tokens and pushed transactions outside the normal withdrawal flow. Binance says it rebuilt the affected systems and closed the entry points, but the lesson lands hard for anyone running crypto infrastructure: human vigilance, a hardened supply chain and ruthless access control are not optional extras, they are the whole game.

Binance reveals how fake job interviews and a poisoned AI tool led to a $6.43 billion crypto heist

#hacking #cybersecurity #data-breach #supply-chain-attack #crypto #social-engineering #binance

View detail →

Researchers at Google Threat Intelligence Group and Mandiant have exposed a fast-growing cybercrime scheme called LLMJacking: criminals break into corporate cloud accounts and resell stolen access to premium AI models on underground marketplaces, leaving the victim to pay the bill. In some cases the charges climbed past $100,000 in a single day and reached up to $1 million before the attackers were cut off.

The playbook is brutally simple. Attackers harvest API keys from public GitHub repositories, abuse misconfigured service account keys or trick users into pasting active session tokens into fake LLM community portals. Then they flip AI services back on and start monetizing quotas through token proxying, automated model streaming and GPU provisioning, all while defenders are still asleep at the wheel.

Services like Jet Night advertise unlimited access to the most advanced models for around $60 a month, a price that only adds up with stolen keys, and one feed actor pushed more than 500,000 model calls in under eight hours. The takeaway for security teams is clear: watch for silent changes to AI accounts, rotate credentials aggressively, and treat exposed API tokens as the front door to a very expensive breach.

Stolen AI keys are turning company cloud bills into million-dollar disasters

#ai #cybersecurity #data-breach #cybercrime #cloud-security #llm

View detail →