OpenAI Says It Appears to Have Been Attacked

OpenAI is struggling to keep its ChatGPT service online.

Denial of Service

OpenAI is struggling to keep its ChatGPT service online.

Outages have been plaguing users of the popular AI chatbot for the past day or so. Users are reporting not being able to login, while others are greeted with the message that "ChatGPT is at capacity right now."

In its most recent update, posted late Wednesday evening, OpenAI blamed a possible cyberattack for the "periodic outages," claiming they were triggered by an "abnormal traffic pattern reflective of a DDoS attack," referring to a distributed denial-of-service attack, which is when hackers target a server by overwhelming it with a flood of traffic.

That's the last time we've heard from OpenAI regarding the outages and according to Downdetector, problems with the company's services are still ongoing at the time of writing.

It's still unclear what the exact cause of the outages was. According to Telegram messages reviewed by TechCrunch, a hacktivist group called Anonymous Sudan, which is linked to pro-Russian hacktivist collective Killnet, took credit.

At the end of the day, though, the purported hack highlights the kind of persistent infrastructural vulnerabilities that plague even the largest and most cutting-edge players on the internet: even if you're developing world-changing AI, it can clearly be hard to keep a website operational day-to-day.

Exceeding Expectations

As far as a motive goes, a member of the hacktivist group posted on Telegram that it was targeting ChatGPT due to it having a "general biasness [sic] towards Israel and against Palestine," per Forbes.

OpenAI CEO Sam Altman initially blamed the outages on renewed interest following the company's Monday announcement of new features including GPT-4 Turbo, its most powerful AI model yet, as well as tools allowing users to create personalized "GPT" AI models.

"Usage of our new features from devday is far outpacing our expectations," Altman tweeted Wednesday afternoon, roughly ten hours before the company blamed a possible DDoS attack. "There will likely be service instability in the short term due to load."

Interestingly, it wasn't just OpenAI's ChatGPT that experienced outages on Wednesday. As CNBC reports, Claude 2, a chatbot run by OpenAI's competitor Anthropic, also experienced issues.

We still don't know if the two instances are linked to each other, but the timing is certainly suspect.

Altman has yet to comment on the possible cyberattack — but a little transparency could go a long way, given the popularity of the company's chatbot.

More on ChatGPT: AI Expert Says ChatGPT Is Way Stupider Than People Realize