Over the previous few years, Uber has schemed to increase its self-driving efforts by spying on rivals, poaching workers, and buying their software program, in accordance to newly launched court docket paperwork. Though competitor intelligence work is commonplace amongst giant corporations, the small print are hardly ever made public.
In the long-running Waymo v. Uber lawsuit, Uber stands accused of stealing and utilizing commerce secrets and techniques from Alphabet’s self-driving automotive division to increase its personal, youthful program. New paperwork filed in federal court docket this week focus on the actions of Uber’s Strategic Services Group (SSG), an eight-person group inside the firm’s Threat Operation division, devoted to amassing intelligence on rivals. A former ThreatOps worker has claimed that SSG often engaged in fraud and theft, and employed third-party distributors to receive unauthorized knowledge or info.
Uber denies these allegations, arguing that they had been a part of that worker’s try to extort cash from the corporate amidst the Waymo lawsuit. But it made SSG workers out there for deposition by Waymo in December. These depositions element the group’s actions, together with video surveillance of rivals’ vehicles, speaking with suppliers, plans to purchase self-driving software program by scraping web sites, and a visit to Las Vegas for the CES commerce present.
The group’s efforts included launching a mission referred to as Zoo to be taught extra about Uber’s self-driving rivals, every of which received its personal code identify. “SSG’s 2017 research will focus on Giraffe, Turtle, Zebra and Turtle/Chimp as well as competitors from Asia,” mentioned a planning doc SSG workers wrote in December 2016, which a lawyer for Waymo learn into the court docket report.
Giraffe was Uber’s code-name for Google and Waymo, court docket information point out. Turtle/Chimp might refer to General Motors and both Lyft (with which it desires to construct a self-driving automotive community) or Cruise, the startup GM purchased in 2016. Zebra—whose purpose is quoted as being to “reinvent completely the automobile”—may be Zoox, a stealthy Silicon Valley startup that’s constructing a robotic taxi from the bottom up.
“Giraffe leads the other 30-plus companies in the race to field fully autonomous vehicles,” admitted the report. “For 2017, SSG’s priority effort will be Giraffe.” SSG was through which automakers Waymo was working with, what know-how wants it was outsourcing, and who its suppliers had been, the information reveal. SSG was additionally tasked with “[tracking] social media accounts, e.g., LinkedIn, Facebook, Twitter, et cetera, in order to map the personal and professional networks of key personnel” at Waymo.
The group’s prime goal was the stuff that makes Waymo’s vehicles so sensible: their secret supply code. “Success in [the] autonomous vehicle race ultimately hinges on the source code,” learn one other SSG doc. “All the source code necessary for success can be compressed to [around] 75 megabytes.” In comparability, an hour-long Netflix obtain is about 200 megabytes.
SSG began its search by inspecting GitHub, an internet site the place software program builders submit all types of open supply software program. “We would [be] looking for context that would be… inadvertently dropped out of there by an engineer [but] we’ve never run across anything that I would consider protected data,” Matt Henley, Uber’s director of ThreatOps, mentioned in a deposition.
The subsequent plan, court docket information present, concerned sending SSG workers to CES in January 2017—autonomous driving is a significant matter on the monumental know-how commerce present in Las Vegas. Team members attended technical shows to see if they might glean any details about different corporations’ supply code that might assist Uber’s Advanced Technology Group construct its personal robocars. Unsurprisingly, that didn’t work both, in all probability as a result of most corporations keep away from revealing key technical knowledge in public lectures.
“So in terms of the number one priorities for collection and organizations, it’s your testimony that Uber obtained no information at CES 2017?… It was a failure?” a Waymo lawyer requested SSG member Edward Russo throughout his deposition. “You could characterize it that way,” Russo replied.
SSG didn’t go away Las Vegas utterly empty handed, the depositions point out. A South Korean firm referred to as Jungsang informed one member of the SSG workforce that it was working with Waymo on laser-ranging lidar methods.
On February 14, 2017, SSG introduced this intel into a gathering with Anthony Levandowski and Lior Ron, ex-Google engineers and co-founders of the self-driving truck startup Otto that Uber had purchased the earlier summer season. Levandowski had been made head of Uber’s self-driving program. He thought the data was “garbage and not of interest,” SSG workforce member Nick Gicinto mentioned in his deposition.
Instead of ready for info to drop into Uber’s lap, Levandowski and Ron steered that the intelligence group take a extra proactive method, in accordance to the brand new materials filed in court docket. Ron famous that offer chain members are “good sources of insight into a competitor’s plans, intentions and capabilities.” A report from the February assembly mentioned, “To succeed, our [intelligence] collection plan must be both broad and deep. It must be broad enough to develop streams of reporting on at least six different competitors… [and] deep enough to acquire in a timely manner meaningful technical data.”
The depositions describe a brand new precedence checklist for SSG, together with exterior distributors, to develop new sources of data, debrief engineers who had left their applications to work at Uber, and “map the human terrain around each of the key personalities” at Waymo.
Shortly after the February assembly, Waymo filed its lawsuit in opposition to Uber, accusing Levandowski of taking 9,700 megabytes of confidential technical self-driving automotive knowledge with him when he left Google, utilizing it to begin Otto, after which bringing it to Uber.
In April, mentioned Gicinto, Russo and one other SSG team-member, Jake Nocon, flew to Phoenix to perform surveillance of Waymo’s experimental self-driving automobiles. For about 4 days, they adopted and videotaped the vehicles driving on public streets. Henley will need to have joined them, as he testified to a dialog in a automotive with Nocon, the place Nocon requested him if there was something to the Waymo lawsuit. “I answered something along the lines, I think they’re just trying to screw with us,” mentioned Henley. “There may be something with Anthony [Levandowski]. And I said, I hope this audio doesn’t leave the car.” But Nocon had unintentionally left the audio recording on, and Waymo discovered it throughout authorized discovery for the trial.
The lawsuit, after all, prompted greater than idle hypothesis amongst Uber staffers. Levandowski and Ron’s plans to method suppliers and conduct every day background analysis on rivals had been by no means carried out, in accordance to a number of SSG depositions. Nick Gicinto testified that he was not conscious of Uber ever hiring exterior distributors to dig up info on Waymo.
Uber in the end fired Levandowski in May 2017 for not cooperating with its investigation into the lawsuit. Gicinto mentioned in his deposition that the brand new head of the autonomous driving program, Eric Meyhofer, shortly requested Ron to stop all surveillance of Uber’s rivals. SSG had additionally filmed different automobiles, together with these run by an unnamed firm in San Francisco, as late as May.
Waymo argues that Uber’s explicit curiosity within the efficiency of its self-driving vehicles final 12 months proves Uber’s in poor health intent. A lawyer for Uber says that “gathering public information on competitors is a common (and legal) business practice,” and that Waymo is making “unwarranted inferences based on irrelevant innuendo.”
The trial is due to start on February 5, at which level, the jury will get to make its personal judgment.