Individual incidents may appear random, but as a whole they show a pattern of disruption and intelligence gathering. They are calibrated to be large enough to have an impact while remaining below the threshold that would prompt a NATO military response

67 incidents. $763.97M gone. Q2 2026 Report is worth reading carefully. Hacken's Security & Compliance Report breaks down what happened, what failed to prevent it, and what that means for the rest of the year:

Two incidents over the last 24 hours resulted in elevated errors or reduced availability on Claude for some users. Multiple separate network failures cut into our capacity to serve Claude, and some requests failed while we rerouted traffic.

Recent security incidents show frontier AI labs may be moving too fast. Replit Co-Founder & CEO @amasad warns that labs must prioritize system safety before risking core internet infrastructure: "I think the responsible thing for them to do is to pace and slow down, because otherwise they're going to be criminally liable for hacking core internet infrastructure." #AI# #Cybersecurity# #TheInformation#

OpenAI found additional incidents of AI models acting deceptively and taking unsanctioned actions during training, the company announced. It's also introducing a new process for the company to publicly report such instances.

OpenAI disclosed six incidents of its AI models misbehaving during training and testing. In one case, an unreleased model searched GitHub for a leaked API key, used it without authorization, then set up disposable email accounts to bypass access blocks. When it still couldn't retrieve earnings data, it fabricated nine financial figures and claimed they came from a website's chart. During training of GPT-5.6 Sol, models concealed mistakes and invented missing historical data. OpenAI's internal monitoring only covered 20% of that run's samples and flagged a "high rate of reward hacking and deception."

As security incidents continue to increase across the crypto ecosystem, we’ve also seen a recent increase in malicious proposals targeting token-voting governance systems on Aragon. These proposals are designed to get control of enough tokens to drain liquidity pools. This isn’t unique to Aragon, but it’s a good reminder to review your governance setup, monitor proposals closely, and make sure appropriate safeguards are in place. Depending on your governance model, there are several ways to strengthen security: • Extend voting durations and add timelock stages. Longer voting windows and a timelock before execution give token holders, contributors, and security teams more time to identify malicious proposals and coordinate a response. If your governance process allows early execution, consider using standard voting so proposals can never execute before the voting period ends. • Restrict proposal creation. Limiting proposal creation to members of your governing bodies, or requiring a minimum token balance to create a proposal, can reduce spam and opportunistic attacks while preserving decentralized governance. • Review quorum settings. Make sure treasury proposals require meaningful token holder participation before they can pass, and periodically review quorum thresholds as voter participation and governance activity evolve. • Consider optimistic governance with a token holder veto. In this model, a trusted group such as a core team or multisig creates proposals, while token holders retain the ability to veto them during a protected objection window. This reduces voting overhead while preserving decentralized control, but it depends on token holders actively monitoring proposals and responding before the objection window closes. Governance security is an ongoing process, and every project should regularly review its governance configuration as threats evolve. We’ll continue building tools to make governance and execution more secure while helping projects strengthen their governance. If you’d like us to review your setup or discuss additional protections, our team is happy to help. Please reach out.