The Washington Post - 05.09.2019

(Axel Boer) #1

THURSDAY, SEPTEMBER 5 , 2019. THE WASHINGTON POST EZ RE A


BY TONY ROMM,
GREG BENSINGER
AND CRAIG TIMBERG

Google will pay $170 million to
settle allegations that it illegally
collected data about children
younger than 13 who watched toy
videos and television shows on
YouTube, settling a long-running
government investigation but
leaving some in Washington once
again furious that regulators had
been outmatched by Silicon
Valley.
In a sweeping complaint, state
and federal regulators alleged
Google knew that some channels
on YouTube were popular among
young viewers and tracked kids’
viewing habits for the purpose of
serving them targeted ads, ulti-
mately raking in “close to
$50 million” from just a short list
of channels that violated federal
children’s privacy laws.
The settlement — brokered by
the Federal Trade Commission
and the attorney general of New
York — brushed aside years of
claims by YouTube that it was not
intended for children under 13 by
citing numerous examples in
which the company bragged to
toymakers such as Mattel and
Hasbro about its popularity
among children. In one boast cit-
ed by regulators, YouTube
claimed to be watched by 93 per-
cent of “tweens,” roughly in line
with independent surveys of the
viewing habits of preteens.
State and federal regulators
said the settlement would force
YouTube and other social media
companies to be mindful when
children under age 13 are using
their services. But a wide array of
privacy advocates, lawmakers and
even some o f the F TC’s o wn mem-
bers said the penalties against
Google would serve as no deter-
rent given its massive revenue.
While the fine set a record for a
violation of the Children’s Online
Privacy Protection Act (COPPA),
the federal law that forbids the
tracking of users u nder 13 without
parental permission, it amounted
to less than two days’ worth of
profits for the tech giant, advo-
cates noted.
The settlement also held no
Google executives directly ac-
countable, didn’t r equire the com-
pany to admit guilt and left some


experts fearful that it might con-
tain a loophole allowing YouTube
to avoid liability for kids’ privacy
missteps in the future.
“It’s good that the F TC is finally
attempting to hold Google ac-
countable for violating COPPA at
a massive scale for a number of
years. But the settlement shifts far
too much of the responsibility
onto the content creators and not
enough to YouTube, a nd the f ine is
woefully inadequate,” said Josh
Golin, executive director of the
Campaign for a Commercial Free
Childhood, an advocacy group
based in Boston that was among
those filing the complaint last
year.
The tension over the settlement
underscored the limits of COPPA,
which i s at o nce one of the nation’s
few federal privacy laws but also,
critics say, o utdated and unevenly
enforced. The FTC has initiated a
process to update the rules en-
forcing the law, and a major re-
write has been proposed by Sen.
Edward J. Markey (D-Mass.), one
of the law’s original authors, and
Sen. Josh Hawley (R-Mo.), but
even their bipartisan proposal has
not yet advanced on Capitol Hill.
“I do not believe that the mo-
ment is slipping away on the pro-
tection of children,” Markey said
in an interview.
The settlement accepts many of

the claims raised by more than 20
consumer and privacy groups in
an April 2018 complaint and
echoes years o f concerns that You-
Tube had become the most popu-
lar site for children while side-

stepping COPPA. Privacy advo-
cates long had pointed out that
many popular channels on the
site are directed to younger audi-
ences, featuring nursery rhymes
as well as videos of children un-
wrapping toys. The Post first re-
ported on the FTC investigation in
June.
The FTC and r egulators in New
York agreed, finding that Google’s
own content-rating systems had
identified certain YouTube chan-
nels as directed toward children.
COPPA has two key tests in
assessing whether a site must
comply: whether it is directed
toward children and whether the

sit e has “actual knowledge” that
children use it. The FTC settle-
ment asserted Wednesday that
YouTube failed both tests after
years of claiming that its site was
not intended for children and

hence not covered by COPPA.
“YouTube touted its popularity
with children to prospective cor-
porate clients,” FTC Chairman Joe
Simons said in a statement. “Ye t
when it came to complying with
COPPA, the company refused to
acknowledge that portions of its
platform were clearly directed to
kids. There’s no excuse for You-
Tube’s violations of the law.”
The settlement was notable for
the extent to which regulators
used YouTube’s own statements
and actions against it. They cited
that YouTube’s own rating system
often labeled videos as intended
for viewers under 13 but still al-

lowed tracking and targeted ad-
vertising. Regulators cited the
channels featuring My L ittle Pony,
Barbie, Thomas & Friends and
Hot Wheels as evidence that they
were intended for a young audi-
ence.
YouTube created a separate app
for children in 2015, called You-
Tube Kids, in part to address c om-
plaints that the main platform
had too many young viewers. But
regulators noted t hat many videos
appeared on both services. The
only difference was that viewers
were not tracked and targeted on
YouTube Kids but were when the
same children’s videos appeared
on YouTube itself.
One popular channel that has
content appearing on both ser-
vices, called Sandaroo Kids, says
on its YouTube page, “We love
dressing in Disney Princess Cos-
tumes, playing pranks and teach-
ing kids how to learn colors,” ac-
cording to the complaint from
regulators. The channel includes
videos with such titles as “Barbie
& Ken Dolls Fashion Show Party
with Doll Ambulance.” The chan-
nel Little Baby Bum, meanwhile,
had videos such as “Bath Song”
and “New Baby Brother & Sister.”
Along with its fine, YouTube
agreed to develop a system that
would require content creators to
indicate whether their videos are
directed toward kids, turning off
tracking for targeted ads in cases
where it believes a child is watch-
ing. YouTube also announced ad-
ditional changes on its own, in-
cluding work to train its software
to detect content that is
kid-oriented using a type of artifi-
cial intelligence known as ma-
chine learning.
“This action is changing You-
Tube’s business model,” said An-
drew Smith, the director of the
FTC’s Bureau of Consumer Pro-
tection. “YouTube cannot bury its
head in the sand.”
But the settlement stopped
well short of what advocates were
seeking. Backed by the agency’s
three Republicans, the settlement
sparked criticism from the FTC’s
two Democrats, who voted
against the agreement even as
they criticized YouTube for profit-
ing improperly from its younger
users.
In a dissenting statement,
Commissioner Rebecca Kelly
Slaughter e xpressed fears that the
government’s settlement requires
too little from YouTube to find and
discipline content creators that
intentionally mislabel their vid-
eos, resulting in a “fence but one
with only three sides.”
Democratic Commissioner Ro-

hit Chopra, meanwhile, faulted
the FTC for failing to obtain a
larger fine for Google “illegally
harvesting children’s d ata,” which
he wrote was “extremely lucra-
tive” for Google. His dissent in-
cludes redacted data suggesting
that behavioral ads improperly
displayed on videos viewed by
kids should have resulted in a fine
in the billions of dollars.
Democrats on Capitol Hill also
blasted the FTC. “When compa-
nies like Google repeatedly break
the law, the FTC must demand
structural change and executive
accountability,” Sen. Richard Blu-
menthal (Conn.) said in a state-
ment. “I am concerned that the
divided vote reflects a lack of
resolve and a lost opportunity to
impose necessary accountability
measures to rein in Google’s pat-
tern of privacy abuses.”
In a rare move, the FTC filed the
settlement in federal court on its
own. Ty pically, such consent or-
ders are filed on the agency’s be-
half by the Justice Department,
suggesting the government may
have been split over s ome element
of the YouTube investigation.
“Following a careful review by its
Consumer Protection Branch, the
Department of Justice declined
the FTC’s proposed settlement,” a
Justice Department official said
Wednesday, who spoke on the
condition of anonymity because
the person was not authorized to
discuss it publicly.
The FTC settlement with
Google marks the latest effort to
probe a nd penalize tech giants f or
their privacy abuses, punish-
ments t hat have served as a litmus
test — even within the agency
itself — about its power to police
Silicon Valley.
In July, the commission issued
a historic, $5 billion fine against
Facebook for allegedly deceiving
users about t he way it collects and
shares their personal informa-
tion. But the penalty is far less
than what some Democrats at the
FTC, lawmakers on Capitol Hill
and privacy hawks had hoped for,
arguing the fine was too small —
and the remedies too weak — to
force significant changes in Face-
book’s business practices.
Earlier this y ear, the FTC issued
a $5.7 million fine against the app
now known as TikTok over charg-
es that it illegally collected data
from children. Democrats chafed
at the FTC’s decision at the time
not to hold specific executives
responsible for the company’s
abuses.
[email protected]
[email protected]
[email protected]

Google to pay $170 million to settle child data allegations


BY DREW HARWELL

Thieves used voice-mimicking
software to imitate a company
executive’s speech and dupe his
subordinate into sending hun-
dreds of thousands of dollars to a
secret account, the company’s
insurer said, in a remarkable case
that some researchers are calling
one of the world’s first publicly
reported artificial-intelligence
heists.
The managing director of a
British energy company, believ-
ing his boss was on the phone,
followed orders one Friday after-
noon in March to wire more than
$240,000 to an account in Hun-
gary, said representatives from
the French insurance giant Euler
Hermes, which declined to name
the company.
The request was “rather
strange,” the director noted later
in an email, but the voice was so
lifelike that he felt he had no
choice but to comply. The insurer,
whose case was first reported by
the Wall Street Journal, provided
new details on the theft to The
Washington Post on Wednesday,
including an email from the em-
ployee tricked by what the insur-
er is referring to internally as “the
false Johannes.”
Now being developed by a
wide variety of Silicon Valley
titans and AI start-ups, such
voice-synthesis software can copy
the rhythms and intonations of a
person’s voice and be used to
produce convincing speech. Te ch
giants such as Google and smaller
firms such as the “ultrarealistic
voice cloning” start-up Lyrebird
have helped refine the resulting
fakes and made the tools more
widely available free, with unlim-
ited use.
But the synthetic audio and
AI-generated videos, known as


“deepfakes,” have fueled growing
anxieties over how the new tech-
nologies can erode public trust,
empower criminals and make
traditional communication —
from business deals and family
phone calls to presidential cam-
paigns — that much more vulner-
able to computerized manipula-
tion.
“Criminals are going to use
whatever tools enable them to
achieve their objectives cheap-
est,” s aid Andrew Grotto, a fellow
at Stanford University’s Cyber
Policy Center and a former senior
director for cybersecurity policy
at the White House during the
Obama and Trump administra-
tions.
“This is a technology that
would have sounded exotic in the
extreme 10 years ago, now being
well within the range of any lay
criminal who’s got creativity to
spare,” Grotto added.
Developers of the technology
have pointed to its positive uses,
saying it can help humanize auto-
mated phone systems and help
mute people speak again. But its
unregulated growth has also
sparked concern over its poten-
tial for fraud, targeted hacks and
cybercrime.
Researchers at the cybersecu-
rity firm Symantec said they have
found at least three cases of
executives’ voices being mim-
icked to swindle companies. The
company declined to name the
victims or say whether the Euler
Hermes case was one of them, but
it noted that the losses in one of
the cases totaled in the millions
of dollars.
The systems work by process-
ing a person’s voice and breaking
it down into components, like
sounds or syllables, that can then
be rearranged to form new phras-
es with similar speech patterns,
pitch and tone. The insurer did
not know which software was
used, but a number of the systems
are freely offered on the Web and
require little sophistication,
speech data or computing power.
Lyrebird, for instance, adver-
tises the “most realistic artificial

voices in the world” and allows
anyone to create a voice-mimick-
ing “vocal avatar” by uploading at
least a minute of real-world
speech.
The company, which did not
respond to requests for comment,
has defended releasing the soft-
ware widely, saying it will help
acclimate people to the new reali-
ty of a fast-improving and “inevi-
table” t echnology “so that society
can adapt.” In an ethics state-
ment, the company wrote, “Imag-
ine that we had decided not to
release this technology at all.
Others would develop it and who
knows if their intentions would
be as sincere as ours.”
Saurabh Shintre, a senior re-
searcher who studies such “ad-
versarial attacks” in Symantec’s
California-based research lab,
said the audio-generating tech-
nology has in recent years seen
“transformative” progress be-
cause of breakthroughs in how
the algorithms process data and
compute results. The amount of
recorded speech needed to train
the voice-impersonating tools to
produce compelling mimicries,
he said, is also shrinking rapidly.
The technology is imperfect,
and some of the faked voices
wouldn’t fool a listener in a
“calm, collected environment,”
Shintre said. But in some cases,
thieves have employed methods
to explain the quirks away, saying
the fake audio’s b ackground nois-
es, glitchy sounds or delayed
responses are actually due to the
speaker being in an elevator, in a
car or in a rush to the next flight.
Beyond the technology’s capa-
bilities, the thieves have also
depended on age-old scam tactics
to boost their effectiveness, using
time pressures, like an impend-
ing deadline, or social pressures,
like a desire to appease the boss,
to make the listener move past
any doubts. In some cases, crimi-
nals have targeted the financial
gatekeepers in company account-
ing or budget departments,
knowing they may have the capa-
bility to send the money instantly.
“When you create a stressful

situation like this for the victim,
their ability to question them-
selves for a second — ‘Wait, what
the hell is going on, why is the
CEO calling me?’ — goes away,
and that lets them get away with
it,” Shintre said.
Euler Hermes representatives
said the company, a U.K.-based
subsidiary of a German energy
firm, contacted law enforcement
but has yet to name any potential
suspects. The insurer, which sells
policies to businesses covering
fraud and cybercrime, said it is
covering the company’s full
claim.
The victim director was first
called late one Friday afternoon
in March, and the voice demand-
ed he urgently wire money to a
supplier in Hungary to help save
the company in late-payment
fines. The fake executive referred
to the director by name and sent
the financial details over email.
The director and his boss had
spoken directly a number of

times, said Euler Hermes spokes-
woman Antje Wolters, who noted
that the call was not recorded.
“The software was able to imitate
the voice, and not only the voice:
the tonality, the punctuation, the
German accent,” s he said.
After the thieves made a sec-
ond request, the director grew
suspicious and called his boss
directly. Then the thieves called
back, unraveling the ruse: The
fake “ ‘Johannes’ was demanding
to speak to me whilst I was still
on the phone to the real Jo-
hannes!” the director wrote in an
email the insurer shared with
The Post.
The money, totaling 220,
euros, was funneled through ac-
counts in Hungary and Mexico
before being scattered elsewhere,
Euler Hermes representatives
said. No suspects have been
named, the insurer said, and the
money has disappeared.
AI developers are working to
build systems that can detect and

combat fake audio, but the voice-
mimicking technology is evolv-
ing rapidly. Google, for instance,
has invested in research and
funded challenges to automati-
cally recognize “spoofed” s peech.
But the company has also devel-
oped some of the world’s most
persuasive voice AI, including
with its Duplex service, which
can call restaurants to book a
table using a lifelike, computer-
generated voice.
“There’s a tension in the com-
mercial space between wanting
to make the best product and
considering the bad applications
that product could have,” said
Charlotte Stanton, director of the
Silicon Valley office of the think
tank Carnegie Endowment for
International Peace. “Research-
ers need to be more cautious as
they release technology as power-
ful as voice-synthesis technology
because clearly it’s at a point
where it can be misused.”
[email protected]

AI software mimics boss’s voice,


gets executive to wire gobs of money


Critics say fine is lenient
for company, which did
not have to admit guilt

‘False Johannes’ may be


harbinger of cyber-scams


as technology improves


ANDREW HARNIK/ASSOCIATED PRESS
Federal Trade Commission Chairman Joe Simons departs Wednesday after announcing that Google
has been fined in a settlement over allegations that it illegally collected data from children younger
than 13 on YouTube. Sen. Richard Blumenthal (D-Conn.) called the settlement “a lost opportunity.”

ALEJANDRO CEGARRA/BLOOMBERG NEWS

A worker carries a ladder to a tree last month during an apple harvest in
Cuauhtemoc, in northern Mexico’s Chihuahua state. The state is Mexico’s most
important apple region, with this year’s harvest expected to be up substantially.

Amid the plentiful harvest


“The settlement shifts far too much of the


responsibility onto the content creators and not


enough to YouTube, and the fine is woefully


inadequate.”
Josh Golin, executive director of the Campaign for a Commercial Free
Childhood
Free download pdf