The phrasing here is key. Musk is not promising that Teslas will become fully autonomous in August, only that there will be activation of a subset of features that will eventually add up to full autonomy. And while cars produced since October of 2016 have all the hardware Tesla says is necessary for self-driving, enabling that hardware requires paying a total of $8,000 in optional fees for many Tesla vehicles.

That issue is better in latest Autopilot software rolling out now & fully fixed in August update as part of our long-awaited Tesla Version 9. To date, Autopilot resources have rightly focused entirely on safety. With V9, we will begin to enable full self-driving features.

— Elon Musk (@elonmusk) June 10, 2018

It’s too early to say what those specific features might be, but the announcement comes as a bit of a surprise, given the dinged reputation of Tesla’s self-driving tech at the moment. Despite its name, Tesla’s existing Autopilot technology is not true self-driving capability, but a set of advanced safety features that have also helped Tesla gather data for training its self-driving software. Autopilot has been involved in a handful of wrecks lately, including at least two in which preliminary reports suggest Autopilot itself may have been at fault.

Get Data Sheet, Fortune’s technology newsletter.

Media coverage of the most recent crash triggered a weekslong Twitter crusade against the press by Musk, who argued that coverage of Autopilot-linked wrecks ignored the larger fact that such features were already safer than unassisted driving. The negative coverage of Autopilot, combined with ongoing challenges with Model 3 production, battered Tesla’s stock, though it has recovered some losses this month.

Setting a firm date for the start of the self-driving rollout suggests Musk remains very confident in his tech—but his careful phrasing in the offhand tweet leaves a huge escape clause. Individual self-driving “features” may amount to little more than further enhancements of Autopilot. There’s still no clear deadline for when a Tesla will be able to fulfill the company’s promise that the cars will take trips “with no action required by the person in the driver’s seat.”

And continuing a piecemeal, step-by-step transition towards full autonomy, while a good way to build on steady progress, could actually increase Tesla’s crash-related headline risk. That’s because, according to many experts, autonomous features are riskiest when human drivers have to maintain attention even while they don’t have full control—that is, with mid-capability systems. Several Tesla crashes have, according to both the company and government investigators, been caused by drivers placing too much faith in Autopilot, taking their hands off the steering wheel and their eyes off the road. Continually touting further self-driving ‘features’, without actually activating full autonomy—even while reducing the intensity of Autopilot attention reminders—might be inviting more of the same.