SShortSingh.
TopTechnologyAI & MLStartupsProgrammingScienceBusinessFinanceCrypto & Web3World
Back to feed
Programming

GMO probes suspected Onamae cloud desktop breach; link to Artex AI tool unconfirmed

0
·2 views

GMO Internet is investigating a suspected breach of its Onamae cloud desktop service, which could affect up to 27,000 servers. Customers have reported unauthorized activity on trading accounts, and attackers have demanded about 8 billion yen in extortion. GMO has shut down the service, notified police, and begun collecting customer devices to search for malware. Reports linking the incident to Artex, a Chinese AI-enabled penetration testing tool, have not been confirmed, and the scale of actual access and data theft remains unverified.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

How to Set Up SSH Key Authentication for Passwordless Remote Server Access

An operations engineer explains how to replace password logins with SSH key authentication, which the author says takes about five minutes using only the command line. The guide recommends generating an Ed25519 key pair with ssh-keygen, with a passphrase to encrypt the private key, and falling back to a 4096-bit RSA key for legacy systems. Public keys can be installed on a server with ssh-copy-id or by appending them to authorized_keys manually. The article also covers testing the new login, disabling password authentication in sshd_config, and troubleshooting common errors like permission problems and agent failures.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Six unpatched Linux kernel flaws disclosed by ZDI, one with public root exploit and no CVE

The Zero Day Initiative reported six Linux kernel vulnerabilities to maintainers and public mailing lists on October 20 and 21, 2026. Sysdig's Threat Research Team found that one already had a working public proof of concept, and several lacked fixes in any released kernel. None had a CVE identifier, so patching teams must track them by ZDI candidate numbers instead of standard vulnerability feeds. TuxCare confirmed the public exploit yields a root shell on several AlmaLinux, CloudLinux, Debian and Ubuntu releases. Sysdig recommends prioritising by the components a fleet actually runs, such as unloading the ksmbd SMB server where unused and restricting port 445.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

Testing Affordable Logging APIs for Small Projects Using Three Silent-Run Scenarios

The article argues that small projects should judge a logging API by whether it can reconstruct a silent scheduled import, not by its lowest price. It recommends emitting four standard events (started, completed, empty, failed) with identifiers, timestamps, result counts, and correlation IDs while keeping customer message content out of routine logs. The author proposes testing three failure fixtures (a skipped run, an empty successful run, and a failed parse) across candidate providers and rejecting any option that cannot reconstruct the last three runs from searchable records. A comparison covers Datadog, Sentry, Better Stack, and Infrai, weighing their fit, limitations, and retention behavior, and includes an example client that keeps the provider origin in configuration so it stays replaceable.

0 comments Read more at DEV Community
0
ProgrammingDEV Community ·

GhostShell: a local terminal tool that remembers past bugs and gates every command

GhostShell is a local-first terminal companion that helps investigate errors by reading a pasted traceback alongside a chosen project directory. It produces a structured analysis with ranked hypotheses, supporting evidence, unverified areas, a risk level, and a follow-up check. The tool stores incidents on the local machine and flags when a similar earlier incident exists, without assuming a past fix will apply again. Suggested commands are displayed for approval and never run automatically, and a local policy check can still block them. Inference runs through a loopback Ollama server, with no cloud model client or telemetry, and the project was built for a Hacktoberfest 2026 AI challenge.

0 comments Read more at DEV Community